Académique Documents
Professionnel Documents
Culture Documents
Edited by
Emmanuelle Labeau and Inès Saddour
ISBN: 978-90-420-3430-3
E-Book ISBN: 978-94-012-0718-8
©Editions Rodopi B.V., Amsterdam - New York, NY 2012
Printed in The Netherlands
Contents
Emmanuelle LABEAU
Aston University
Inès SADDOUR
Aston University
Over the last twenty five years, the study of time, aspect and – to a lesser
extent – mood acquisition has enjoyed increasing popularity and a constant
widening of its scope. In such a teeming field, what can be the contribution of
this book? We believe that it is unique in several respects. First, this volume
encompasses studies from different theoretical frameworks: functionalism vs
generativism or function-based vs form-based approaches. It also brings
together various sub-fields (first and second language acquisition, child and
adult acquisition, bilingualism) that tend to evolve in parallel rather than
learn from each other. A further originality is that it focuses on a wide range
of typologically different languages, and features less studied languages such
as Korean and Bulgarian. Finally, the book gathers some well-established
scholars, young researchers, and even research students, in a rich inter-
generational exchange, that ensures the survival but also the renewal and the
refreshment of the discipline.
The first part of the volume is devoted to the study of child language
acquisition in monolingual, impaired and bilingual acquisition, while the
second part focuses on adult learners. In this section, we will provide an
overview of each chapter.
The first study by Aviya Hacohen explores the acquisition of composi-
tional telicity in Hebrew L1. Her psycholinguistic approach contributes
valuable data to refine theoretical accounts. Through an innovating
methodology, she gathers information from adults and children on the
influence of definiteness, number, and the mass vs countable distinction on
the constitution of a telic interpretation of the verb phrase. She notices that
the notion of definiteness is mastered by children as young as 10, while the
mass/count distinction does not appear before 10;7. However, this does not
entail an adult-like use of telicity. She therefore concludes that beyond
definiteness and noun type, pragmatics may play an important role in the
derivation of Hebrew compositional telicity.
For the second chapter we move from a Semitic language to a Slavic
one. Milena Kuehnast focuses on the acquisition of negative imperatives in
Bulgarian, a form that presents the specificity of being grammatical only with
the imperfective form of the verb. The study examines how 40 Bulgarian
children distributed in two age-groups (15 between 2;11-3;11, and 25
between 4;00 and 5;00) develop with respect to the acquisition of
imperfective viewpoints, and the use of imperfective morphology. It shows
an evolution in the recourse to expression of force in the use of negative
imperatives, as well as the influence of morphological complexity on the
successful production of forms.
With Yi-An Lin’s study, we concentrate both on another type of
informant and of framework. Indeed, he studies the production of children
suffering from Specific Language Impairment (SLI), a developmental
language disorder the causes of which include cognitive impairment, psycho-
emotional disturbance, and motor-articulatory disorders. Using the Leonard
corpus in CLAN, Lin aims to test two competing accounts of SLI (the
Agreement and Tense Omission Model [ATOM] and his own Phonetic Form
Deficit Model [PFDM]) that conflicts on the role attributed to spellout in the
impairment. Spellout is the point at which the Computational System for
Human Language (CHL) passes over the most recently derived part of the
derivation to the interface components, Phonetic Form (PF) and Logical
Form (LF). ATOM claims that SLI sufferers have a deficit in their syntactic
representation while PFDM suggests that the problem only occurs at the
spellout level. After studying the corpus from the point of view of tense /
agreement marking, case marking, argument-movement and auxiliary
inversion, Lin finds further support for his model.
Olga Gupol, Susan Rohstein and Sharon Armon-Lotem’s chapter offers
a welcome bridge between child language acquisition and multilingualism.
Their study explores the influence of intensive exposure to L2 Hebrew on the
development of L1 Russian tense and aspect morphology through an elicited
narrative. Their informants are 40 Russian-Hebrew sequential bilingual
children distributed in two age groups 4;0 – 4;11 and 7;0 – 8;0. They come to
the conclusion that bilingual children anchor their narratives in perfective like
monolinguals. However, while aware of grammatical aspect, bilinguals lack
the full form-function mapping and tend to overgeneralize the imperfective
on the principles of simplicity (as imperfective are the least morphologically
marked forms), universality (as it covers more functions) and interference.
Rafael Salaberry opens the second section on foreign language learners.
In his contribution, he reflects on the difficulty L2 learners of Spanish
encounter when it comes to distinguishing between iterativity (conveyed with
the use of the preterite) and habituality (expressed through the imperfect). He
Introduction iii
examines in turn the theoretical views that see, on the one hand, habituality
as part of grammatical knowledge and iterativity as pragmatic knowledge,
and on the other hand both habituality and iterativity as grammatical
knowledge. He comes to the conclusion that the use of preterite as a default
past tense marker may explain the impoverished system of aspectual
distinctions, not only at beginners but also at advanced levels, which may
indicate that the system is differentially represented among L1 and L2
speakers. Acquiring the vast array of functions conveyed by a form is
therefore no mean feat, as confirmed by the next study.
Based on the prototype theory, Kathleen Bardovi-Harlig’s chapter
focuses on the development of the progressive in L2 English. It opens with an
overview of the functions of the progressive in English. Then, a review of
acquisition research on the progressive in English and other languages is
provided. The bulk of the chapter reports on a longitudinal study of 16
learners of L2 English and shows how their use of the progressive expands
from the prototypical uses of process and continuousness to the less
prototypical uses of repetition and future. The study concludes that the
progressive spreads in interlanguage in accordance with prototype accounts.
However, it suggests additional stages, not predicted by the Aspect
Hypothesis, in the development from activities and accomplishments at least
for the meaning of repeatedness.
A similar theoretical framework is adopted in the following chapter, but
it deals with a lesser studied language. Hyun-Jin Kim revisits the claims of
the Aspect Hypothesis in relation to the acquisition of L2 Korean by two L1
English learners. Inspired by studies on L2 Japanese, she focuses on the
emergence and spread of the past / perfective marker –ess- and the
progressive – ko iss- in the interlanguage of her informants throughout their
third and fourth semesters of study. The data collected through six sessions of
conversational interviews and picture description tasks seem to support the
Aspect Hypothesis. Indeed learners show a strong association between past
tense and accomplishments / achievements at the start and a gradual
extension to other types; a limited use of past / perfective marker with states
and an affinity of progressive with activities / accomplishments and later
achievements. In addition, – ko iss– moves from progressive to resultative in
the specific category of Korean verbs meaning wear / carry.
While the previous contributions focus on function, Evgeniya Sergeeva
and Jean-Pierre Chevrot’s is interested in form. The authors explore the
acquisition of verbal morphology in L2 French by 30 instructed native
speakers of Russian distributed in a low and high levels. They use an
elicitation task for verbs with different models of stem alternation and study
how token frequency and base forms influence stem selection. The analysis
shows that frequency affects correct production, especially among learners
with high proficiency. As for substitution errors, it appears that forms with a
iv Emmanuelle Labeau & Inès Saddour
simple structure are systematically more frequent than the target form they
replace. When a complex form serves as a substitute, it is more frequent only
when it is replacing another complex form. As regards the use of base forms,
the 3rd person singular of the present – and to some extent the infinitive –
play this role in the corpus. The authors therefore conclude that the
processing of surface forms can be influenced positively or negatively by the
frequency of the target forms and of other competing stems, and by the
proximity of the target stem to a base form.
Finally, Martin Howard’s contribution takes up the challenge of
focusing on the poorer relation of the TAM system. On the basis of L2
French data obtained through sociolinguistic interviews, he studies the
expression of futurity, conditional and subjunctive in three groups of
university learners with classroom teaching only (two or three years of
university teaching) or with a mixture of classroom teaching and naturalistic
exposure (2 years at University + 1 year abroad). An analysis of relative
frequencies leads him to suggest a continuum of use going from futurate
present to conditional with past hypothetic conditional clauses in si, which
needs to be confirmed by further studies.
Acknowledgements
Aviya Hacohen
Ben-Gurion University of the Negev
1. Introduction
In Hebrew however, this test does not distinguish telic from atelic predicates:
(2) a. sara ratsa kilometer tox Sa'a / bemeSex Sa'a
Sara run-SG.F.PAST kilometer in hour / for hour
Sara ran a kilometer in an hour / for an hour
b. sara ratsa tox Sa'a / bemeSex Sa'a
Sara run-SG.F.PAST in hour / for hour
Sara ran in an hour / for an hour
Instead, one of the contexts that does distinguish telic and atelic predicates in
Hebrew is the modifier kim'at ('almost') (Smith, 1991):
(3) a. sara kima't tsav'a bad
Sara almost paint-3SG.F.PAST material
Sara almost painted material
b. sara kim'at tsav'a ribua
Sara almost paint-3SG.F.PAST square
Sara almost painted a square
As can be seen above, (3a) has only one interpretation, namely, that Sara did
not paint at all. Conversely, (3b) is ambiguous: one reading is that Sara did
not paint at all, as in (3a), but it may also mean that Sara did start painting but
did not finish. This ambiguity is the result of the heterogeneous nature of telic
predicates as opposed to the homogeneity that characterizes atelic predicates
(cf. the next section): telic predicates involve both a process and a result
(Smith, 1991). Since kim'at ‘almost’ may modify either the process or the
result, the two possible readings above emerge.
Another diagnostic of telicity has been termed ‘the imperfective
paradox’ (Dowty, 1979). The examples in (4) illustrate the different
entailments associated with telic and atelic predicates when they are
generated in the progressive:
These examples show that the atelic predicate in (4a) entails the perfective,
while the telic predicate in (4b) does not. Though Hebrew does not encode
the progressive morphologically, using the inflected preposition beod
(‘while’) provides the desired context (Yitzhaki, 2003):
(5) a. be'odo melatef xatul hadelet niftexa
while-3SG.M stroke-3SG.M.PRES cat the-door open-3SG.F.PAST
While he was stroking a cat, the door opened
b. be'oda metsajeret igul haiparon niSbar
while-3SG.F draw-3SG.F.PRES circle the-pencil break-3SG.M.PAST
While she was drawing a circle, the pencil broke
Thus, the telic and the atelic predicates exhibit different behaviors: while the
sentence in (5a), using an atelic predicate, entails that the cat was stroked, the
sentence in (5b), containing a telic predicate, does not entail that a circle was
drawn.
Having shown this, let us now turn to a survey of the crosslinguistic
theoretical literature on compositional telicity.
2. Theoretical background
2.1. Defining (compositional) telicity
1
This relationship between the part and whole events in telicity described here as
homogeneity, has been otherwise analyzed in terms of a homomorphic relation
existing between the event and the object (see Krifka, 1992).
4 Aviya Hacohen
paint the square completely (cf. Tenny, 1994). It is precisely this property of
telic predicates that will be tested in the experiments, as will be shown later.
Crucially, the notion of homogeneity refers to both the verbal and the
nominal part of the predicate. That is, the «homogeneity value», and
consequently the telicity value of the predicate is the result of the
combination of the properties of the verb and those of the direct object NP.
The view that the direct internal argument plays a crucial role in the
derivation of telicity is widely accepted in the theoretical literature (e.g.
Verkuyl, 1972; 2005; Dowty, 1979; Krifka, 1989; 1992; 1998; Tenny, 1994).
Specifically, the telicity value of the predicate is ultimately determined by
whether the direct object NP is quantized (e.g. Verkuyl, 1972, Krifka, 1989,
Tenny, 1994). Quantization is taken to be motivated by one of two nominal
properties of the direct object argument, namely, noun-type and/or
definiteness. These effects are illustrated by the Hebrew data below:
(7) a. litsboa bad (mass-atelic)
paint-INF material
To paint cloth
b. litsboa ribua (count-telic)
paint-INF square
To paint (a) square
(8) a. litsboa ribuim (indefinite-atelic)
paint-INF squares
To paint squares
b. litsboa et haribuim (definite-telic)
paint-INF ‘et’ the-squares
To paint the squares
c. litsboa et habad (definite mass-telic)
paint-INF ‘et’ the-material
To paint the material
As can be seen from the examples above, similarly to English, it is the nature
of the NP which determines the telicity value of the predicate. Specifically, a
quantized NP combined with an incremental dynamic verb derives a telic
predicate.
Over the past three decades, a number of studies have shown that
crosslinguistically, children’s initial use of the past tense is restricted to telic
predicates while the present or imperfective are used primarily with atelic
predicates (Bloom, Lifter & Hafitz, 1980; Shirai & Andersen, 1995 for
English; Bronckart & Sinclair, 1973 for French; Stephany, 1981 for Greek;
Antinucci & Miller, 1976 for Italian; Aksu-Koç, 1988 for Turkish; Berman,
On the acquisition of compositional telicity in Hebrew 5
1983 for Hebrew). This (production) pattern, which became known as the
Aspect First Hypothesis, has lead researchers to argue that tense morphology
in early child language serves to encode aspect.
The work of Angeliek van Hout (e.g. 1998) has provided important
comprehension data. Using the Truth Value Judgment task (Crain & McKee,
1985; Crain & Thornton, 1998), van Hout tested Dutch and English speaking
adults and typically developing (TD) children. Four verb-types were used:
intransitives, transitives with a bare object, particle verbs, and transitives with
an object preceded by a possessive pronoun. Two verbs were used, eat and
drink, and all items were presented as both complete and incomplete event-
types. Of most interest for the current study is the incomplete condition, and
in particular, items that include transitive verbs preceded by a possessive
pronoun. Sample excerpts from the Dutch and English protocols in the
transitive condition are presented below:
(9) [Dutch incomplete event]
EXP: Hier is een witte muis. Hij heeft een net stuk kaas gevonden. Kijk, hier is
hij aan het eten. Hij knabbelt er een beetje af, maar dit stuk is veel te
groot voor hem. Hij laat nog wat over voor later.
[English incomplete event]
EXP: Here’s a white mouse. He just found a piece of cheese. Look, here he’s
eating. He takes a couple of bites, but his cheese is too big for him for
now. He leaves a piece for later.
The story was accompanied by three pictures depicting the beginning of the
event, the event in progress, and the final result. Having heard the story
above, the subject is asked a yes / no question about the character. An
example of such a question is given in (10):
(10) Dutch: Heeft de witte muis zijn kaasje gegeten? Heeft de rode muis zijn
kaasje gegeten?
English: Did the white mouse eat his cheese? Did the red mouse eat his
cheese?
Three groups of 15 Dutch speaking children (3- 4- and 5-year olds) and one
group of 15 adults participated in the Dutch experiment. In the English
experiment there were 19 3-year-olds, 17 4-year-olds, 11 5-year-olds, and 16
adults, all of whom were speakers of American-English.
Van Hout found that Dutch speaking adults rejected the predicate 80%
of the time, which strongly indicates that this predicate has a completion
entailment in adult Dutch 2. Performance of 4- and 5-year-olds was at chance,
2
But note that these data are for only two predicates, eat cheese and drink coke,
which does not necessarily generalize over to other telic predicates.
6 Aviya Hacohen
and rejection rate of the youngest children was 20%. Rejection rate of
English speaking adults was only 25% and three child groups behaved at
chance. These results, particularly the adult data, go against the predictions,
and could ostensibly suggest that the telic / atelic distinction is not a
psychologically real phenomenon of adult English. However, it is quite clear
that this is not the case; rather, it seems that the surprising results are due to
flaws in the methodology. The first problem is in the choice of structure. The
possessive pronoun, zijn / his, rather than the definite article, yields a telic
reading in Dutch but not in English. Secondly, the choice of verbs: though
eat and drink are perhaps the most commonly used verbs in the theoretical
literature, psycholinguistic data, as well as results from my pilot studies, have
shown that adult speakers' judgments of the telicity value of predicates with
these verbs varies significantly (e.g. Ogiela, Casby and Schmitt, 2005).
Given that English speaking adults did not clearly reject predicted telic
predicates as descriptions of incomplete events, it follows that it is impossible
to say anything meaningful about the acquisition of compositional telicity in
English.
Based on van Hout’s (1998) experiment, Schulz & Penner (2002) tested
the acquisition of compositional telicity in German. Materials used in this
experiment were very similar to the ones used by van Hout: three picture-
sequences depicting an eating / drinking event accompanied by a story.
Following this presentation, the subject was asked two yes / no questions
about the event, one intransitive question and one question using a particle
verb or a transitive verb frame. The latter being the relevant one for my
purposes. An example of a transitive question is given below in (11):
(11) Hat das Mädchen den Apfel gegessen?
has the girl the apple eaten
Did the girl eat the apple?
Each subject was asked two such questions in the incomplete condition. One
group of 24 German-speaking adults and a group of 24 German-speaking
children aged 4;1-6;4 (mean 5;4) were tested. Results show that behavior of
both groups was very near chance (56% 'yes' response for adults and 52% for
children). Thus, as in van Hout’s study, the German experiment did not yield
the predicted results even for adult language.
More recently, Wagner & Carey (2003) argue that English-speaking
children aged 3-5 are sensitive to the telic / atelic distinction. The children
were asked to count how many times an event depicted in a film occurred,
Furthermore, van Hout (p.c.) reports that when she (and others) tried to replicate
this result, with similar versions of the experiment, rejection rates were much
lower, suggesting that the completion entailment was not nearly as strong as the
original experiment suggests.
On the acquisition of compositional telicity in Hebrew 7
the event description was either telic or atelic (e.g. paint a flower / paint).
Sample stimulus questions appear in (12).
(12) Telic description
How many times does the girl paint the flower?
Atelic description
How many times does the girl paint?
(14) Yesterday Julie got up early. It was her son's birthday. She usually liked to
surprise him for his birthday. She decided to surprise him with a birthday
cake.
Julie made a cake.
Julie made cakes.
Results from the aspectual interpretation task revealed that while Spanish
speakers demonstrated native-like knowledge of (a)telicity, Bulgarian
speakers did not. Interestingly, though, the Bulgarian group did perform
similarly to the other groups on the atelic sentences but not on the telic
sentences. Data from the stories task corroborated this finding, with the
Bulgarian learners showing significantly more inaccuracy on the telic stories
than the atelic ones. This effect was not found for the Spanish learners.
Thus, low intermediate level Spanish speaking learners of English were
highly accurate in their interpretation of (a)telicity in English, while L1
Bulgarians were only accurate with respect to atelic structures.
The current project aims to discover when Hebrew-speaking children
master telicity. The hypotheses and predictions which guide this study are
presented in the following section.
As for child language, recall that child results from van Hout (1998) and
Schulz & Penner (2002) are not very different from the adult data obtained.
This being the case, and taken together with the possible methodological
flaws of the experimental design, it would be wrong to base hypotheses and
predictions for TD Hebrew speaking children on those experiments. Instead,
based on the Aspect First Hypothesis, according to which the lexical-
semantic notion of telicity is present cross-linguistically from the initial
emergence of verbal forms, it is hypothesized that
3
Though, naturally, it was not possible to have a perfect correlation, since not
many verbs allow all three NP types as the direct object argument. Specifically,
the mass condition seems to be the most restricted one in this context, and thus,
the verbs in the two mass conditions are quite different to those in the other four
conditions.
10 Aviya Hacohen
Definiteness
+ -
5 items 5 items
Singular Example: Example:
count litsboa et haribua litsboa ribua
color-INF 'et' the-square color-INF (a) square
5 items 5 items
NP type
Participants were then presented with video-clips showing the relevant events
and had to judge whether the accompanying (a)telic predicate, which was
orally expressed by the experimenter, matched the event or not. An example
from the definite singular count condition is given in (16) below:
On the acquisition of compositional telicity in Hebrew 11
It is important to note that the visual stimuli were kept constant across all
items in terms of “incompleteness” of the event depicted, while the telicity
value of the verbal stimuli was manipulated across conditions. In other
words, all the events depicted in the videos were incomplete, i.e. not reaching
their natural endpoint. The main reason for using only incomplete scenarios
is that it was important to have at least five different verbs in each condition,
so that a generalization could be made based on more than two verbs. In an
ideal situation, both the complete and the incomplete scenarios would have
been used; however, given the number of different predicates, and the fact
that I wanted all participants to judge all the predicates, having both complete
and incomplete counterparts for each predicate would have made the
experiment too repetitive and much too long. Furthermore, though using only
incomplete scenarios is clearly a compromise, it nevertheless provides the
two crucial contexts required for a psycholinguistic task, namely, acceptance
(when atelic predicates are used in the verbal stimulus) and rejection (when
telic predicates are used).
5.1.2. Participants
role, the results of all the children were collapsed. Participants were recruited
from two kibbutzim, as well as a major city and the geographic area covered
the north and the south of Israel. All participants were tested individually by
the author.
Rejections were coded as 0 and acceptance was coded as 1. Adult results are
presented in figure 1 below:
100%
91%
87%
90% definite
80% indefinite
70%
acceptance rate
60%
50%
40%
27%
30%
20%
11% 11%
10% 4%
0%
singular plural mass
NP TYPE
4
It is beyond the scope of the current paper to discuss this unexpected result, but
see Hacohen (in preparation) for a discussion of this issue.
On the acquisition of compositional telicity in Hebrew 13
with very high acceptance rates for both the indefinite plural (87%) and the
indefinite mass (91%). It is important to note that acceptance rates would
have been even higher (95% and 98% respectively) if not for one subject who
had exceptionally high rejection rates in these two conditions.
Child data are presented in Figure 2 and 3 below.
100%
90% definite
80% indefinite
70% 65%
62% 62%
acceptance rate
60%
50% 45%
40%
30% 26%
20%
20%
10%
0%
singular plural mass
NP TYPE
Let us first survey the conditions that were predicted to be telic, i.e. rejected.
As can be seen in Figure 2 above, the younger group had an acceptance rate
of 26% in the definite singular condition, and 20% in the indefinite singular.
In the definite plural condition, acceptance rate was 40% and in the definite
mass, these children had an acceptance rate of 62%. For the two atelic
conditions, namely the indefinite plural and the indefinite mass, acceptance
was at 62% and 65% respectively.
The main analysis involved Age Group (child, adult) as a between-
subject variable and Condition as a within-subject variable. A main effect
was found for Condition (F(5,100)=43.98, p<.0001) but not for Age Group
(F(1,20)=.72, p=.41). However, crucially, the 2-way interaction was
significant (F(5,100)=11.63, p<.0001).
A series of planned comparisons looked at the differences between the
age groups in each of the conditions separately. The difference between
14 Aviya Hacohen
children and adults was significant (or marginally significant, in one case) in
all conditions, except for the two singular count conditions. F(1,20)=1.82,
p=.19 for the definite singular count; F(1,20)=.95, p=.34 for the indefinite
singular count; F(1,20)=9.78, p<.01 for the definite plural, F(1,20)=3.91,
p=.06 for the indefinite plural; F(1,20)=5.68, p<.05 for the definite mass;
F(1,20)=5.74, p<.05 for the indefinite mass.
Finally, an analysis was conducted comparing predicates that were
predicted to be telic ((in)definite singular, definite plural and definite mass)
with those that were predicted to be atelic (indefinite plural and mass), across
the age groups. A main effect was found for Telicity (F(1,20)=207.94,
p<.0001), but not for Age Group (F(1,20)=.003, p=.96). Most importantly,
the 2-way interaction was significant (F(1,20)=52.56, p<.0001), showing
that the effect of Telicity was much greater for the adults than for the
children. The effect for the children was nonetheless clearly significant
(F(1,12)=50.00, p<.0001), revealing a sensitivity to the telicity value of the
verbal stimulus. In sum, data from this experiment confirm the predictions
for adult Hebrew, with subjects rarely accepting telic predicates as true
descriptions of incomplete events and regularly accepting atelic predicates as
descriptions of the same incomplete events. Furthermore, it was shown that
indeed definiteness and the mass / count distinction play a crucial role in the
derivation of telicity in adult Hebrew.
The predictions for child Hebrew were clearly not confirmed, as
Hebrew speaking school-age children evinced high acceptance rates for telic
predicates as descriptions of incomplete events, alongside a relatively low
acceptance rate for atelic predicates as descriptions of the same events. The
data do reveal that the children in this experiment were sensitive to the telic /
atelic distinction; however, it is very clear that Hebrew-speaking children are
non-adultlike in their knowledge of telicity even at the age of 12 5.
But what causes the children’s non-adultlike behavior? Data from the
current study showed that the two main factors determining the telicity value
of predicates in adult Hebrew are definiteness and noun type. It follows, then,
that in order to establish the source of the non-adultlike behavior on telicity,
each of these elements must be tested independently of telicity. This was
done using two auxiliary experiments, as described in the following two
sections.
5
In this sense, results of this experiment are comparable to what Wagner & Carey
(2003) found for (younger) English speaking children.
On the acquisition of compositional telicity in Hebrew 15
6. Experiment 2: Definiteness
6.1. Method
6.1.1. Design and procedure
6
Balaban has been developing the original questionnaire as part of a larger
project, which investigates the pragmatic system of Hebrew speaking adults
suffering from brain damage to the right-hemisphere.
16 Aviya Hacohen
6.1.2. Participants
Adult and child results are presented in Figure 3 below. As can be seen from
the graph, both adults and children rarely accepted the use of definite NPs in
non-unique contexts (10% and 15% respectively). Unique contexts always
licensed the use of definite NPs, with acceptance rates of 100% for all the
subjects.
Since no variability was found in the control condition, only the
experimental condition, the ‘Definite in non-unique’, was analyzed. No
significant difference was found between the age groups, F(12) =.48, p =.64.
Thus, data from this experiment revealed that 9-12 year old Hebrew-speaking
children have adultlike knowledge of the appropriate use of definiteness.
Once it was shown that definiteness cannot be the source of the non-adultlike
On the acquisition of compositional telicity in Hebrew 17
100%
90% children
adults
80%
70%
acceptance rate
60%
50%
40%
30%
20%
10%
0%
Definite in non-unique Definite in unique
NP TYPE
7. Experiment 2: Definiteness
7.1. Method
7.1.1. Design and procedure
Using Barner & Snedeker’s (2005) Quantity Judgment Task, five different
noun types were examined, using five experimental conditions: count nouns
(e.g. efronot ‘pencils’), substance-mass nouns (e.g. kemax ‘flour’), flexible-
count nouns (e.g. niyarot ‘papers’), flexible-mass nouns (e.g. niyar ‘paper’),
and object-mass nouns (e.g. do'ar ‘mail’). There were four items per
condition and 12 filler items, all randomly ordered. As illustrated in (18), for
each item, participants were presented with two characters, one with two
large objects and the other with five small objects of the same kind. The
smaller number of items always constituted the greater volume of material.
The verbal stimulus, lemi yesh yoter X? (‘who has more X?’), was the same
across items and items were all randomly arranged.
18 Aviya Hacohen
It is important to note that, as can be seen from the examples above, in the
two flexible conditions (the flexible-count and the flexible-mass) the visual
stimulus remains constant while the verbal stimulus, i.e. count and mass
syntax, is manipulated.
7.1.2. Participants
The data were coded such that responses based on individuation, i.e. judging
the character with the larger number of items as having ‘more', were given a
score of 1. The opposite response, i.e. judging the character with the more
overall volume as having ‘more’, was given a score of 0. The results are
presented in Figure 4 below:
90% count
flexible count
80% object mass
mass
70% flexible mass
60%
50%
38%
40%
30%
20%
11%
10%
0%
0%
4;4-6;1 7;9-10;3 10;7-12;0 Adults
AGE GROUP
As can be seen from the graph, adult Hebrew speakers behaved according to
the predictions. They always based their judgments on the number of
individual items in the two count conditions and in the object-mass condition
(at 100% for all three conditions). Conversely, they only very rarely based
their judgments on number in the two mass conditions (0% for the substance-
mass condition and 4% for the flexible-mass condition). In the child data, a
clear developmental pattern emerges. Children in the youngest age group
On the acquisition of compositional telicity in Hebrew 21
References
Milena KUEHNAST
Centre for General Linguistics (ZAS), Berlin
1. Introduction 1
1
I’m grateful to Emmanuelle Labeau and two anonymous reviewers for their
suggestions and insightful comments on an earlier version of the paper.
2
Bulgarian grammar books identify 9 tenses (possible with both aspects), 3 of
which are expressed synthetically and 6 analytically. See Bojadzhiev et al.
(1999) for a systematic description. For a brief comparison with the tense
systems of the other Slavic languages, see Sussex & Cubberley (2006:242ff.).
3
In Bulgarian, imperfectivisation is a grammaticalised means for the expression
of historical present. See S.Guentcheva’s (1981) interval-based explanation for
the retained processual meaning and the actual present uses of imperfectivised
verbs of different derivational types.
4
The absolute ungrammaticality of synthetic NI with perfective verbs in
Bulgarian is employed as a test for the aspectual value of a verb (Bojadzhiev et
al. 1999:490). In other Slavic languages, the aspect of the verb produces two
different interpretations of NI – a prohibitive one with imperfective verbs and a
preventive one with perfective verbs (cf. Chrakovskij & Volodin 1986).
Acquisition of negative imperatives in Bulgarian 29
5
The multi-functional particle da, formally equivalent to the consecutive
conjunction that, expresses imperative force.
30 Milena Kuehnast
(3) a. pusn stem PF -a 1SG PRES IND - pusn stem PF -i 2SG IMP
‘to drop – drop!’
b. nale stem PF -ja 1SG PRES IND - nale stem PF -j 2SG IMP (-te 2PL)
‘to pour – pour!’
6
There are also positive imperative utterances demanding that the existing
situation is preserved. In the next section it will be argued that negative
imperatives belong to the type [-change]. Compare also Birjulin's (1994:48-60)
classification of imperatives according to the feature [+/- required change of the
existing situation.]
7
Note that Bulgarian encodes nominal definiteness by means of enclitic definite
articles. In the given example, the bare noun sok would acquire a type reading as
in (1); the use of the definite article –a in soka “the juice” marks a specific
quantity of juice.
Acquisition of negative imperatives in Bulgarian 33
S Requested Situation:
an empty jug becomes a full jug
OT
Requested situation:
S
Preserve: empty jug becomes full jug
OT
In utterance (9) the speaker wants the addressee to preserve the initial state of
affairs, namely to leave the jug empty. The reference time of the imperative
(OT) is included in the situation time, yielding an imperfective interpretation.
Under negation, we obtain a combination of temporal values known as
imperfective present – both the time of speaking and the source state of the
situation denoted by the verb are included in the reference time of the
obligation.
34 Milena Kuehnast
In Slavic languages, the present tense form of perfective verbs does not
convey the meaning of actual present. When used as predicates in main
clauses, bare perfectives express future tense, which is their regular temporal
function. This is not the case in Bulgarian, where future tense is expressed
analytically by a future auxiliary. The use of the future auxiliary is equally
obligatory with perfective and imperfective verbs. Therefore in Bulgarian,
there are no negative synthetic imperatives with perfective verbs, which are
regularly employed in preventive utterances by speakers of the West and East
Slavic languages (Kuehnast 2008) 8.
Bulgarian verbs offer intricate ways of depicting and viewing the internal
temporal constitution of events. Perfective aspect is encoded by means of
perfective prefixes and one suffix, while imperfective aspect is marked by
imperfective suffixes. Both aspects can also be expressed by non-derived
verbs, primary perfectives and imperfectives. There is an ongoing discussion
of the nature of Slavic type aspect, mainly on the question of whether the
morphological paradigms of perfective and imperfective verbs indeed
represent viewpoint aspect (cf. Smith, 1991, for a discussion of Russian
aspect) or whether real aspectual distinctions are encoded consistently only
8
In the cited paper I argue that the unavailability of perfective synthetic NI is due
to the impoverished temporal functions of bare perfective verbs and also to the
basic principle in the organisation of the TAM system of Bulgarian concerning
overtly differentiated and precise form to function mappings. For a different
approach see D. Levinson (2005), who relates the phenomenon of NI to
propertites of nominal case. Thanks to an anonymous reviewer for drawing my
attention to Levinson’s work.
Acquisition of negative imperatives in Bulgarian 35
for the past tenses (Bertinetto 2001). In Slavic aspectology, the debate
revolves around the distinction between purely aspectual affixes, which alter
the aspectual value of the verb without changing its lexical meaning, and
derivational affixes, which combine derivational and grammatical functions.
For the purposes of the present study we will concentrate on
imperfective suffixes and the process of imperfectivisation in Bulgarian.
Imperfective verbs derived by means of imperfective suffixes do not differ
from their perfective counterparts in lexical content, but only in aspectual
value (Maslov 1981, Bojadziev et al. 1999). Therefore, imperfectivisation is
understood as a grammatical process. Imperfectivisation is a means of
viewpoint change and is operative in the aspectual systems of all Slavic
languages. Nevertheless, there are differences in the degree of application
and in the grammatical functions of the resulting imperfective verbs
(Ivančev 1978).
Imperfective derivation by means of imperfective suffixes applies to
almost all perfective verbs in Bulgarian. Primary imperfectivisation applies to
simple perfective verbs. Secondary imperfectivisation applies to derived
perfective verbs 9. Perfective verbs derived from simple imperfectives
through prefixation maintain the prefixed stem (11). Perfective verbs derived
by means of perfective suffixes such as -n receive the imperfective suffix
after stripping the perfective one (12).
(11) lepja primary impf – zalepja prefixed pf – zalepvam secondary impf ‘to glue’
(12) lepja primary impf – lepna suffixed pf – lepvam secondary impf - semelfactive meaning
9
Not all Slavic languages apply secondary imperfectivsation to a similar degree.
For instance in Russian, a secondary imperfective is not available for verbs
derived by means of purely aspectual prefixes. This fact is used to prove the
derivation of true aspectual pairs. In Bulgarian, secondary imperfectivisation
always applies, if not barred on phonological grounds.
36 Milena Kuehnast
3. Empirical investigation
IMPERFECTIVATION PATTERN
PERFECTIVE
Productive: suffix -va Minor: suffix– a / stem change
DERIVATION
3.3 Procedure
The children were tested in a quiet room in the kindergarten. The child and
the two experimenters sat together at a table. The first experimenter
introduced to the child a second experimenter who carried a glove puppet in
the shape of a penguin. She explained that this was the penguin Toto who had
come to Bulgaria from the South Pole. Toto is not familiar with the rules at
the kindergarten. The experimenter explained that she was going to play
some jokes on the penguin, asking Toto to do different things. The child was
asked to reject the silly requests and tell the penguin not to carry them out,
because such behaviour would be inappropriate in the kindergarten. The child
was encouraged to address the penguin directly.
Figure 3. Elicitation task
10
To my knowledge, there is no sufficiently large corpus of Bulgarian which can
be used for frequency counts at the moment.
40 Milena Kuehnast
The procedure has the advantage of making the child direct prohibitive
requests not to an adult person, but to a penguin puppet in a pragmatically
well motivated situation. Stated in this way the task avoids potential
problems with the authority cline between adults and children in a direct
interaction.
Subjects received 2 or 3 training stimuli, and once they were able to
understand the task, the experimental stimuli were given. The stimuli were
written on cards and presented in random order. The sessions were audio
recorded and subsequently transcribed.
There are 3 types of short elliptical utterances (without lexical verbs) which
express different degrees of obligation depending mainly on intonation and
gestures and partly on the lexical means chosen. The imperative force used
varies from prohibition expressed by the simple rejection “No !” and with
proper intonation, to different degrees of permission expressed by simple
negative modals (13) or constituent negation (14).
(13) Exp: Toto, stăpi pf na topkata !
Step on the ball !
Child(2;11): Ne mozhe !
This is not allowed !
(14) Exp: Toto glătni pf kopcheto !
Toto, swallow the button !
Child (2;11): Ne kopcheto n, def !
Not the button !
(15) Exp: Toto, nadraskaj pf lista !
Toto, scribble over the page !
Child (4;02): Da ne nadraskva 3SG PRES impf lista !
(He) should not scribble over the page !
(16) Exp: Toto, skochi pf dolu !
Toto, spring down !
Child (4;05): Njama da skochish 2SG FUT pf !
(You) will not spring !
the target system. In the present case, the youngest subjects seem to reverse
the principle. By means of constituent negation they deny the definite object,
which measures out the event (in the sense of Verkuyl 1993) and arrive at an
atelic interpretation akin to the imperfective viewpoint required in a negative
imperative.
The remaining alternative constructions (15) and (16) impose an
obligation on the addressee more or less directly. The rates of the more
neutral 3rd person analytic da-constructions, which help to avoid addressing
the interlocutor with a direct command remain relatively stable. The use of
periphrastic da-constructions becomes reinforced by the fact that they allow
for perfective verbs. Such utterances are interpreted rather as warnings than
as commands.
Children aged 4;7 - 5;0 do not avoid direct requests, producing 84%
synthetic NI. The only analytic example with an embedded da-construction
expresses the performative and the propositional part of the prohibitive
separately (17).
(17) Exp: Toto, priberi pf igrachkite !
Toto, collect the toys !
Child (4;7): Ne ti davam da gi pribirash 2SG PRES impf !
I don't permit you to collect the toys !
In a few cases the subjects used statements with future predicates. The
truth commitment contained in the tensed predicate expresses degrees of
certainty that the denied situation will not be brought about which are
comparable to those of the synthetic negative imperatives. With age, children
feel more comfortable with the position of authority which is needed in order
to impose on an interlocutor an obligation concerning a counterfactual
situation. With respect to the end effect desired by the speaker, synthetic
negative imperatives are quite subtle expressions. The request to the
addressee to keep the present state of affairs is linguistically marked only
through the imperfective form of the imperative predicate.
100%
% target forms
75%
2;11 - 3;11
50%
4;0 - 5;0
25%
0%
simple prefixed suffixed
Perfective derivation
44 Milena Kuehnast
Older children who have had more time to become acquainted with the target
aspectual system are highly sensitive to the imperfectivisation requirement in
NI and tend to combine imperfectivisation strategies, sometimes producing
hybrid imperfectivised forms (18) – (19).
(18) Exp: Toto, nadraskaj prefixed pf masata!
Toto, scribble over the desk !
Child: Toto, ne ja * draskvaj prefix stripping + suffix –va !
Toto, don't scribble it over !
target: nadraskvaj
(19) Exp: Toto, izleti prefixed pf prez prozoretsa !
Toto, fly away through the window !
Child: Toto, ne *izlitvaj stem change + suffix -va !
Toto, don't fly away !
target: izlitaj
The frequency of such usages effects that the overall imperfective marking of
prefixed perfectives increases, but at the same time the rate of target
imperfectivised forms remains low.
46 Milena Kuehnast
The experimental data shows that the children produced negative imperatives
with imperfective forms reliably. However, the error patterns obtained cannot
be sufficiently accounted for only by considering the properties of the
morphological paradigm. Productivity, transparency and frequency of usage
exert a significant impact on the acquisition process of aspectual
morphology. Nevertheless, the error analysis above demonstrates that
children do not apply the productive imperfectivisation suffix in all cases.
Which factors could possibly strengthen the preference of the younger
children to apply prefix stripping to verbs of the productive class and not to
the verbs of the minor class ? We want to propose that these factors are
related to the interpretation of the prohibitive utterances in terms of the
temporal schema expressed in the lexical item and its representation in the
situation model.
A closer look at the experimental items which are available to prefix
stripping reveals that these verbs correspond to accomplishment and
achievement situation descriptions. The verbs from the productive imper-
fectivisation class denote accomplishments; the verbs from the minor class
denote achievements. Prefix stripping yields the simple imperfective verb
from which the perfective verb is derived. The basic imperfective verb
depicts an atelic activity. In a negative imperative context, the application of
prefix stripping to achievement and accomplishment predicates yields quite
different results. The activity expressed by the resulting simple imperfective
is located either before (20) or after the negated change of state (21).
The irregular prefixed perfective verbs used in the experiment are
derived by means of ingressive suffixes and depict achievement situations.
The punctual change of state applies to the beginning of the activity denoted
by the basic imperfective verb and situates this activity in the target state. In
the case of prohibitive utterances containing an ingressive achievement
predicate, prefix stripping yields an expression which does not fit the
causality chain of the situation model. The use of the simple imperfective
verb appears unfortunate as it depicts a situation referring to the subsequent
activity (21). The fact that prefix stripping was used sporadically is due to the
general-factual meaning of simple imperfective verbs – all phases of the
action are included in the basic denotation. Vinnitskaya and Wexler (2001)
found similar production behaviour in Russian children, who overused
imperfective forms for situations in which adult speakers produced perfective
verbs.
(20) a. Morphological derivation pattern:
draskam simple impf – nadraskam prefixed pf – nadraskvam sec. impf
‘to scribble – to scribble sth. over – to be scribbling sth over ’
Acquisition of negative imperatives in Bulgarian 47
b. Temporal chain:
scribbling (source state) – object scribbled all over (target state)
(21) a. Morphological derivation pattern:
letja simple impf – izletja prefixed pf – izlitam sec. impf
‘to fly – fly away – to be flying away’
b. Temporal chain:
starting to fly (source state) – flying activity (target state)
the simple imperfective verb, thus rejecting the activity of the source state. In
such cases, the children preserved the view of the accomplishment as an
‘indivisible whole’ (Comrie 1976) encoded in the prefixed perfective stem
and did not zoom in on the internal temporal structure of the event.
The obtained results fit well in the ‘perspective-based’ approach to the
acquisition of imperfective aspect, which is grounded in studies exploring the
acquisition of perfective and imperfective past tense meaning to form
mappings. Knowledge of the perfective / imperfective contrast is associated
with the ability of the children to vary their perspective towards the event.
Other lines of reasoning (i.e. lack of pragmatic knowledge restricting the use
of past imperfectives as in Vinnitskaya & Wexler (2001) or problems with
discourse integration as in van Hout (2005)) also draw on past tense
experiments. The complex interrelations of tense and aspect in past tense
utterances and the different discourse-related readings such utterances may
obtain, make it difficult to disentangle the effects of these factors on the
acquisition of imperfectivity as aspectual notion.
The current experiment abstracts away from the discussion revolving
around the imperfective past tense uses and the associated failure / success of
temporal linking to other past events (van Hout 2005; Kazanina & Phillips
2007). The experimental method targets the ability of a child to map his / her
‘here and now’ viewpoint onto the internal structure of the event. The main
findings support the view that young children acquire perspective shifting
only gradually. The significant correlation between appropriate uses of
imperfective morphology and verb types found in the production data
indicates that the conceptual similarities between the properties of some
situation descriptions and the characteristics of an event internal perspective
boosts the acquisition of imperfective aspect.
Although they are successful in imperfectivising primary and suffixed
perfective verbs, Bulgarian children experience difficulties when they have to
shift their perspective to the source state of a two-state predicate and to mark
this shift appropriately up to the age of 5. The experimental data based on the
production of coerced imperfective negative imperatives provides evidence
that 3 to 5-year-old Bulgarian children are still on their way to acquire all
facets of imperfective aspect available in the TAM system of Bulgarian.
References
Yi-An LIN
St. John’s University, Taiwan
1. Introduction
1
I am grateful to my M.A. supervisor Andrew Radford for his insightful
contribution to the development of the Phonetic Form Deficit Model in this
study, as well as in my Master’s dissertation submitted at the University of
Essex.
2
In GB or MP, Case refers to abstract Case, while case refers to morphological
case.
Omission Model (ATOM) and Lin’s (2006) Phonetic Form Deficit Model
(PFDM), are evaluated on the basis of the Leonard corpus of spontaneous
speech production data in the Child Language Data Exchange System
(CHILDES). If SLI involves a syntactic deficit, one should find that syntactic
operations, such as Argument movement (A-movement) and Case
assignment, will be a problem for children with SLI. In contrast, if SLI
involves a phonetic form (PF) deficit, one should find that children with SLI
commit errors on PF operations, such as tense marking, agreement marking
and auxiliary inversion, but leave syntactic operations intact.
2. Theoretical background
2.1. Minimalist programme
According to Chomsky (1995, 2000, 2001, 2004), the main aspects of the
Minimalist Programme (MP) are motivated by general principles underlying
the way in which all biological systems operate, language being just a
representative element of these systems. Language acquisition is determined
by a biologically endowed innate language organ called the Faculty of
Language (FL). Universal Grammar (UG) is the theory of the initial stage of
FL and might be seen as a unified model of the distinguishing features of
human languages.
The architecture of FL outlined in the latest of Chomsky’s papers (1995,
2001, 2004) includes a cognitive system and performance systems. The
cognitive system stores information and makes this information available to
the performance systems that access it in language use. A natural language in
the human cognitive system consists of two basic components: a Lexicon
(Lex) and a Computational System for Human Language (CHL). It is the Lex
that feeds the building blocks of the sentence into the CHL. The Lex
represents a mental dictionary of all substantive and non-substantive lexical
items in the language (LIs), which are characterized by their idiosyncratic
traits. These LIs are seen as feature-bundles of phonological, semantic and
formal features. A particular language selects features from the store of
features made available by UG and is identified by a listing of combinations
of these features in the Lex. In other words, it is in the Lex that categories are
assumed to be specified for the properties that determine the language-
specific, but UG-constrained, syntax. It is the locus of parametric variation.
Therefore, everything that people have to acquire in order to know a
particular language is in the Lex.
The CHL, which performs Narrow Syntax (NS), is a step–by-step
structure-building system which combines primitive elements from Lex into
larger units. The numeration is an array or a selection of LIs taken from the
Lex, and it functions as a pre-syntactic workspace for those selected LIs to be
fed into the CHL to build a syntactic structure. It also indicates how many
To spellout, or not to spellout ? That’s the deficit 57
times LIs are to be used in a structure; therefore, it constitutes the initial point
of the structure-building process. Spellout is the point at which the CHL passes
over the most recently derived part of the derivation to the interface
components, Phonetic Form (PF) and Logical Form (LF). PF is the interface
component that translates the syntactic structure into a format readable by the
Articulatory-Perceptual system (A-P), which computes the structure’s
pronunciation. LF is the interface component that translates the syntactic
structure into a format which the Conceptional-Intentional (C-I) system can
deal with to compute the conceptualisation and interpretation of the structure.
Chomsky (1995) proposes that the computation of an expression (Exp)
converges at an interface level only if Exp consists solely of elements
providing instructions to the external systems that will make use of those
instructions. Features that are legible to the external systems at the interface
level are interpretable, while all other features are defined as uninterpretable.
In other words, only the interpretable features can survive to the LF and PF
representation, and the uninterpretable features must be eliminated. An
uninterpretable feature reaching the interface will cause the derivation to
crash. Both the interpretable and uninterpretable features are attribute-value
pairs. Furthermore, the unvalued features are uninterpretable features, but not
vice versa. In this framework, valued means the feature is given a value upon
selection, and unvalued means the feature must be valued during the
derivation by Agree.
According to Chomsky (2000, 2001), Agree is a head-head relation,
which allows the valuation and erasure of an uninterpretable feature, by
matching it with an identical feature of another item, in a sufficiently local
domain. When a syntactic object (a head) with unvalued features is merged
with another object (its complement), it serves as a Probe which searches for
a matching Goal (a constituent which has identical interpretable features with
which the Probe can agree). Matching of the features of the Probe under
identity with features of the Goal is sufficient to delete the uninterpretable
features on the Probe, rendering movement unnecessary. According to this
conception, Agree is driven by uninterpretable features on the Probe, which
must be deleted for legibility. In other words, the Probe represents the
element which seeks to be determined, while the Goal is the element which
satisfies the Probe. For minimal computation, a Probe should search the
smallest possible domain, namely its c-command domain, to find its Goal.
The relation Agree is established between a Probe and a Goal if there is no
element closer to the Probe than the Goal with the relevant feature values. In
addition, when the Probe bears an Extended Projection Principle [EPP] 3
feature, the movement of the Goal will be triggered.
3
By convention, features are enclosed in square brackets and often abbreviated.
58 Yi-An Lin
The Agreement and Tense Omission Model of SLI originates from Wexler’s
(1994, 1996) studies on clause structure and inflection in young TD children.
The grammar of TD children can be captured in terms of Wexler’s Optional
Infinitive (OI) stage. He claims that TD children undergo a protracted stage
during which they alternate between producing finite and infinitive forms of
verbs in contexts where finite forms are required in adult grammar. In
addition, TD children at the OI stage tend to omit auxiliaries and copula BE 4
in finite contexts. Further, Wexler asserts that optional infinitives arise
because of the underspecification or omission of the tense feature in the
clause representation.
Adopting Wexler’s idea, Rice, Wexler and Cleave (1995) propose that
children with SLI go through an Extended Optional Infinitives (EOI) stage in
which a similar pattern of optional infinitives is found until they are 7 or 8
years of age. Rice and her collaborators propose that the grammar of children
with SLI is characterized in terms of the tense omission model. They predict
that children with SLI will display limited proficiency in the use of
morphemes marking tense, while leaving other inflectional morphemes
unaffected. In addition, when these children do specify the tense feature, they
respect all its morphosyntactic properties. That is, when children with SLI
use tense morphemes or auxiliaries, they use them correctly in the same way
as TD children do. Nevertheless, such a tense-deficit hypothesis wrongly
predicts that children with SLI never use accusative subjects with past tense
verbs.
In later work, Schütze and Wexler (1996) claim that optional infinitives
can result from the underspecification or omission of either the tense or
4
In this chapter, the capitalized word is the label for various forms of that word.
To spellout, or not to spellout ? That’s the deficit 59
For instance, children with SLI may have problems with auxiliary inversion
but leave Wh-movement intact. More specifically, Lin’s model makes the
following predictions:
(i) children with SLI may have problems with tense marking even though the
[TNS] feature is correctly specified on T;
(ii) children with SLI may have problems with agreement marking even
though the [PERS] and [NUM] features (or the [AGR] feature in ATOM) are
correctly specified on T;
(iii) children with SLI may have problems with case marking even though the
[CASE] feature of nominals is correctly specified;
(iv) children with SLI will not face any problems with A-movement and Wh-
movement;
(v) children with SLI may experience difficulties with auxiliary inversion or
DO-support.
3. Methodology
Since ATOM argues that children with SLI may sometimes leave verbs
underspecified for the [TNS] feature in finite contexts, it predicts that the
To spellout, or not to spellout ? That’s the deficit 61
5
SD stands for standard deviation.
62 Yi-An Lin
The corpus data reveals that all the ESLI children frequently use bare
verb forms in contexts where the 3SgPres –s is required. Nine of the children
show difficulties in marking 3SgPres on main lexical verbs. A detailed
summary of each child’s performance with agreement marking on main
lexical verbs is given in Table 2 below:
Table 2. Summary of Each Child’s Agreement Marking on Main Lexical Verbs
Number
Number of obligatory of Percentage use of suffix –s
Child
contexts for suffix –s s-inflected in obligatory contexts
verbs
A 133 54 40.60%
B 49 7 14.29%
C 67 43 64.18%
D 40 13 32.50%
E 35 12 34.29%
F 63 44 69.84%
G 45 14 31.11%
H 81 15 18.52%
I 18 4 22.22%
J 47 1 2.13%
K 23 2 8.70%
Overall 601 209 34.78% (SD=20.74)
The data indicates that six of the ESLI children mark agreement on
auxiliaries below chance level. A detailed summary of each child’s
performance with agreement marking on the auxiliaries is given in Table 3.
As shown in Tables 2 and 3, the mean percentage group scores for the
agreement marking on main verbs and auxiliaries are 34.78% (SD = 20.74)
and 37.11% (SD = 26.52), both below chance level.
Although ATOM correctly predicts the ESLI children’s difficulties with
tense and agreement marking, a theoretical problem with the account arises.
The assumption that the functional head T, which only contains the
uninterpretable [AGR] feature and the interpretable [TNS] feature, can be
underspecified for the [TNS] feature, violates Chomsky’s (1995) claim that
syntactic structures must not contain heads to which no semantic
interpretation can be assigned. Since the interpretable [TNS] feature on T
provides information about the time at which an event takes place, T cannot
be assigned any semantic interpretation if the [TNS] feature is not specified,
which in turn causes the derivation to crash at the semantic interface
(Radford 2005a).
However, given the fact that the ESLI children in question have no
problems with Nom Case assignment, as shown in 4.2, their difficulties with
tense and agreement marking should be attributed to a problem of retrieving
correct verb forms, given Chomsky’s (2000, 2001) assumption that the
unvalued [CASE] feature of subjects is valued as Nom (and is deleted) by a
64 Yi-An Lin
finite [TNS] feature carried by a T head fully specified for the [PERS] and
[NUM] features. In other words, these children’s difficulties with tense and
agreement marking cannot result from a syntactic deficit which causes the
underspecification of the [TNS] and/or [AGR] feature(s).
In contrast, PFDM treats these tense marking and agreement marking
errors as spellout errors. That is to say, these errors result from a problem of
retrieving correct verb forms in the PF component.
6
See Appendix for the raw frequency of personal pronoun forms used by each
child.
To spellout, or not to spellout ? That’s the deficit 65
4.3. A-movement
Although it does not make any specific prediction about ESLI children’s
performance with A-movement, ATOM claims that they frequently
underspecify the [AGR] feature in obligatory contexts, which should result in
frequent A-movement errors, given Chomsky’s (2000, 2001) proposal that
agreement plays a key role in A-movement.
However, the corpus data suggests that ESLI children do not commit
any word order errors in obligatory contexts where A-movement is required.
Instead, the children correctly move subjects to the specifier position of TP
even when they commit other errors, such as tense/agreement marking errors
and auxiliary-omission errors, shown in the sentences below:
(3) It not work (Child B)
(4) He not clean (Child G)
(5) It not working (Child H)
(6) This be not the doctor (Child H)
Although it does not make any specific prediction about ESLI children’s
performance with auxiliary inversion, ATOM maintains that children with
SLI sometimes underspecify [TNS] and/or [AGR] feature(s) on T. This
should lead to problems with auxiliary inversion, since current thinking
within Minimalism suggests that auxiliary inversion is triggered by either an
affix on complementiser (C) carrying a [TNS] feature requiring it to attract
the closest head marked for tense (Pesetsky and Torrego 2001; Radford
2004) or an affix on C carrying a set of strong agreement features requiring it
to attract the closest head specified for agreement features (Radford 2005b).
More specifically, if tense is the factor triggering auxiliary inversion,
children’s performance on auxiliary inversion is expected to reflect their
performance on tense marking. If agreement is the factor triggering inversion,
their performance on auxiliary inversion should have the same frequency as
agreement marking.
It is interesting to explore the extent to which ESLI children’s
performance on auxiliary inversion can be explained in terms of ATOM and
the theories for the mechanism of auxiliary inversion. Examination of the
Leonard corpus data reveals that the overall frequency with which these
children supply auxiliary inversion (52.45%) is more similar to the rate at
which they supply tense (53.87%) than to the rate at which they supply
agreement (34.78%). Furthermore, according to the findings in the previous
sections, the [AGR] feature seems to be unimpaired in the ESLI children’s
grammars. Therefore, their problems with auxiliary inversion seem to result
from the underspecification of the [TNS] feature on C.
However, following Chomsky’s (2001) proposal that auxiliary
inversion, a T-to-C head movement, is a PF operation, PFDM explains the
children’s poor performance with tense marking, agreement marking and
auxiliary inversion in a more principled way. According to PFDM, neither
the [TNS] nor the [AGR] feature is underspecified in the NS. The major
problem for children with SLI lies in spelling out the [TNS] and [AGR]
To spellout, or not to spellout ? That’s the deficit 67
5. Concluding remarks
In this study, two competing accounts of SLI, namely Wexler, Schütze and
Rice’s (1998) Agreement and Tense Omission Model (ATOM) and Lin’s
(2006) Phonetic Form Deficit Model (PFDM), are evaluated on the basis of
the Leonard corpus in CHILDES. In the previous sections, it has been shown
that the ESLI children in the Leonard corpus face difficulties in the use of the
3SgPres morpheme –s and the past tense verb forms, as predicted by ATOM
and PFDM. Furthermore, the data suggests that case marking is unimpaired
in these children’s grammars and that the ESLI children do not have any
problem with A-movement. Both findings suggest that tense and agreement
marking errors are not due to the underspecification of the [TNS] and [AGR]
features on T as proposed by ATOM. If the [TNS] and [AGR] features on T
were indeed underspecified, the assignment of Nom Case should become a
problem, leading to the incorrect case marking on subjects. In addition, the
underspecification of [TNS] and [AGR] features on T should result in the
non-occurrence of A-movement for these ESLI children. However, these are
not the findings, as shown in 4.2 and 4.3. Moreover, given the fact that the
children have problems with auxiliary inversion, as reported in 4.4, it is
suggested that the tense and agreement marking errors are merely spellout
errors, for these three types of errors, namely tense marking, agreement
marking and auxiliary inversion, are all PF operations instead of syntactic
operations.
References
Appendix
Raw Frequency of Personal Pronoun Forms Used by Each Child
Child Type Nom Weak Gen Strong Gen Acc
1Sg I = 125 my = 39 my = 2 me = 30
me = 2
1Pl --- our = 2 --- ---
2 you=53 your = 10 yours = 2 you = 11
3MSg he = 124 his = 18 --- him = 14
A
him = 1 him = 2
3FSg she = 1 her = 1 --- her = 3
3NSg it = 35 its = 2 --- it = 101
3Pl they = 4 --- --- them = 16
them = 6
I = 170 my = 26 mine=7 me = 8
1Sg
me = 2 my = 1
1Pl --- --- --- ---
2 you = 10 --- --- ---
B
3MSg --- his = 2 --- him = 2
3FSg --- her = 1 --- her = 1
3NSg it = 20 --- --- it = 52
3Pl they = 2 --- --- them = 18
I = 203 my = 21 mine = 1 me = 4
1Sg
me = 1
we = 25 our = 2 --- we = 1 7
C 1Pl
us = 4
2 you = 67 your = 1 --- you = 5
3MSg he = 75 his = 45 --- him = 2
7
This item may be a performance error because the child made a self-correction
immediately in the next sentence. Thus, it is excluded from the count of his
performance on case-marking.
70 Yi-An Lin
Olga GUPOL
Susan ROTHSTEIN
Sharon ARMON-LOTEM
Bar Ilan University, Israel
1. Introduction
category of bi-aspectual verbs in Russian that have the same form for both
perfective and imperfective aspects: e.g., kaznit’IMPF / PRF ‘to execute’,
amputirovat’IMPF / PRF ‘to amputate’ and others (Forsyth 1970; Čertkova
1996). This category once constituted 4.1% of the verbal lexicon, but has
since increased due to borrowing(Gagarina 2002a; Ožegov 1989; Čertkova
1996).
The majority of Russian verbs stand in aspectual imperfective-perfective
opposition relationships reflected by verbal morphology, in which one
member is morphologically marked, while the other is unmarked (Borik
2002). The unmarked form is called the base form. Normally, the base form
is the imperfective, although some verbs have an unmarked base form which
is perfective, from which imperfectives can be derived via secondary
imperfectivization (a process which applies also to derived perfective verbs).
It is important to note that Russian morphology allows derivation of multiple
perfective forms from a single imperfective base that can denote rather
different events in comparison with each other and with the base verb as in
examples 1 (a), 1 (b), 1(c) and 1(d):
(1) a. Ivan čitalIMPF / PAST knigu.
Ivan read / was reading a / the book.
b. Ivan počitalPRF / PAST kniguACTIVITY.
Ivan read / has read a book for a while
c. Ivan pročital PRF / PAST kniguACCOMPLISHMENT.
Ivan read / has read a / the book.
d. Ivan dočital PRF / PAST kniguACCOMPLISHMENT.
Ivan read / has read a / the book up (to the end).
The perfective verbs in examples 1.b. (počitat’ PRF ‘to read for a while’), 1.c.
(pročitat’ PRF ‘to read’) and 1.d. (dočitat’ PRF ‘to read up to the end’) are
derived from the imperfective base verb čitat’ IMPF ‘to read’ in 1.a. and stand
in opposite aspectual relations with it: počital PRF / PAST ‘read for a while’ in
1.b denotes a delimited activity (Padučeva 1996, 2008) with an explicit
starting and terminal point, while pročital PRF / PAST ‘read’ in 1.c. and dočital
PRF / PAST
‘read up to the end’ in 1.d. denote completed accomplishments with
obvious terminal boundaries as well. In contrast to both perfective readings,
the morphologically unmarked imperfective counterpart in 1.a. (čital knigu
IMPF / PAST
‘read / was reading a / the book’) does not have explicit terminal
boundaries encoded in the verbal morphology. Nevertheless, 1.a. allows
ongoing, incomplete or completed interpretations as can be seen in 2.a., 2.b.
and 2.c., respectively:
(2) a. Ivan čital IMPF / PAST knigu s dvux do pyati.
Ivan was reading a / the book from two to five.
76 Olga Gupol, Susan Rothstein & Sharon Armon-Lotem
fact that perfective verbs denote sets of events associated with a bounded
measure. The imperfective is associated with a number of unbounded
readings, as mentioned above i.e. iterative, incomplete or interrupted, and
ongoing perspective (i.e. ongoing at the perspective point of the utterance),
where the ongoing event can ultimately be complete or interrupted.
Unlike Russian verbs, Hebrew verbs have 4 typological stem bases,
which differ from one another due to the different position of the guttural
letters in the verbal stem. These stems appear traditionally in seven
derivational conjugations that assign to the same verbal root different lexical
meanings (Berman 1978). Hebrew verbs have subject-verb agreement for:
a) 1st, 2nd or 3rd persons in past and future tenses, unlike Russian,
b) feminine or masculine gender in all three tenses in contrast to
Russian,
c) singular or plural numbers across past, future and present forms.
Unlike Russian, Hebrew verbs have no morphological manifestation of
aspect, which is a lexical feature in Hebrew. The different morphological
representations of the verbal predicates in 1(a), 1(b), 1(c) and 1(d) only have
lexical representations as in example 3:
(3) Ivan karaPAST / TELIC / ATELIC et ha-sefer / sefer.
Ivan was reading / read / read for a while / has read a book.
That is, the inherent temporal properties of verbal predicates are not
lexicalized by means of morphology in Hebrew and can only be derived
compositionally, by means of adverbial modifiers, subordinate or adjacent
clauses, and their combinations. Recent studies on Hebrew aspectuality
provide evidence that Hebrew is sensitive to differences in Vendler’s classes
(Boneh and Doron 2005; Greenberg 2008; Tsarfaty 2004; Yitzchaki 2004).
The above overview of the Russian and Hebrew morphosyntactic
paradigms presents two rather different morphological systems, which are
available to the Russian-Hebrew bilingual and may cause the erroneous use
of verbal morphology in contrastive structures such as:
a) typologically different stem patterns,
b) two inflectional conjugation types in Russian versus five
derivational conjugation types in Hebrew,
c) different distribution of agreement morphology across tenses,
d) different gender systems, i.e. there is no neutral gender in Hebrew,
e) presence versus absence of the morphological marking of aspect.
These typological differences between the languages can thus influence the
acquisition of the verbal morphosyntax in bilingual children's L1 Russian.
Since the main difference between the two languages is the presence versus
The development of L1 Russian tense-aspect morphology 79
monolingual errors during this time. These mistakes gradually decrease with
the increase in the complexity of verbal utterances in monolingual production
(Gagarina 2003).
Production studies of Russian monolinguals’ naturalistic data show that
the knowledge of imperfective and perfective verbal forms and a rule-
governed tense system are evident early on in children's production (Bar-
Shalom and Snyder 2001; Bar-Shalom 2002, 2003; Brun, Avrutin and
Babyonshev 1999; Gagarina 2000, 2002b, 2003, 2004; Gvozdev 1961 among
others). The early stages of tense / aspect acquisition are governed by the
semantic bias. Russian monolinguals show a preference for the use of the
perfective past with the naturally telic lexical classes of accomplishments and
achievements and imperfective present with atelic activity and state lexical
classes (Stoll 1998; Gagarina 2000). The production studies on the early
tense and aspect acquisition show that Russian monolinguals aged 1;6 - 2;11
(Bar-Shalom 2002) and 1;8 - 2;5 (Gagarina 2000, 2003, 2004a, 2004b) show
mastery of tense and aspect morphology. The use of aspectual counterparts
becomes rule governed by that age as we can see in example 4. adapted from
Gagarina (2004a):
(4) a. Mama stroilaPAST / IMPF / ATELIC / ACTIVITY.
Mother was building. (Vanya, 2; 4)
b. Ogromnyj postroilPAST / PRF / TELIC / ACCOMPLISHMENT, ogromnyj dom.
Huge built, huge house. (Vanya, 2; 3)
selective use of tenses, rather than morphological aspect, which does not
exist in Hebrew. The data showed that young Hebrew monolinguals tend to
use telic predicates in the past tense and the atelic predicates in the present
tense. The acquisition of morphological expression of grammatical aspect is
not relevant, since there is no explicit manifestation of grammatical aspect in
Hebrew.
The development of tense and aspect in Russian and Hebrew
monolingual narratives is also similar. Anstatt (2005) and Berman and
Neeman (1994) reported that 80% of Russian monolinguals aged 5-10 and
67.7% of Hebrew monolinguals aged 3-11, respectively, favour anchor past
tense in their narratives (see figures 1 and 2). Russian monolinguals prefer to
tell stories in the perfective past tense (71 %). The present tense is not
favoured by Russian or Hebrew monolinguals in narratives.
Figure 1: Russian monolinguals Figure 2: Hebrew monolinguals (Berman
(Anstatt et al. 2005) et al. 1994)
12,5%
1% 19,8%
19%
9%
71%
61,6%
20% Aspect
0% Stem
errors
I II III IV I II III
When presented with a picture, in which a dog was drawn, the Russian–
Hebrew bilingual child is expected to use the perfective form of the verbal
predicate (narisovalaPAST / PRF ‘drew’), which is an accomplishment and
denotes the completed event with the obvious terminal boundaries. However,
the child uses the morphologically unmarked imperfective counterpart
(risovalaPAST / IMPF ‘was drawing / drew’) of the verb, which does not have
explicit terminal boundaries encoded in the verbal morphology and which is
inappropriate in the context. Since the perfective reading is derived from
imperfective for accomplishments, i.e. the imperfective verb is the base form
for accomplishments in Russian; the child has chosen the imperfective form
risovalaPAST / IMPF ‘was drawing / drew’ as the universal one for this lexical
type of verb. Armon-Lotem et al. (2006) suggest that this is because Hebrew,
which does not have the morphological manifestation of aspect, has only one
translation equivalent for both readings of the verb in Russian. Though the
study reported that Russian-Hebrew bilinguals tend to use mostly
imperfectives instead of perfectives, it was found that they make errors in
both directions as can be seen in examples 7 (a) and 7 (b):
(7) a. Int: Oj! On v dome ne pomeshaetsya. Shto on delaet.
Oj! He does not fit into the house. What is he doing?
Eldar: On pokushal PAST / PRF / ATELIC / ACTIVITY.
He has eaten (for a while).
(Eldar, 5;0; context: describing the picture in which the dinosaur is eating
pancakes)
a’. On kushajet PRES / IMPF / ATELIC / ACTIVITY .
He is eating. (Intended utterance)
86 Olga Gupol, Susan Rothstein & Sharon Armon-Lotem
Figure 4 shows the distribution of aspectual errors for early and late
bilinguals. Notably, early bilinguals use more perfectives instead of
imperfectives than late bilinguals. This developmental trend is typical for
monolingual children at the onset of verb production (Gagarina 2000) and
can thus be attributed to a delay in aspectual acquisition.
Figure 4: Distribution of aspectual errors in late and early Russian-Hebrew bilinguals
(Armon-Lotem et al. 2006)
12
10
8
6
4
2
0
Taking into account the fact that late bilinguals started to acquire their
L2 Hebrew after the age of 3, i.e. after their L1 verbal morphosyntax had
been acquired, the studies claimed that errors found in both Russian-Hebrew
populations indicated a delay for the early bilinguals and attrition for the late
bilinguals in the verbal morphosyntax development. Delay and L2 influence
on the acquisition of the L1 Russian aspectual system in simultaneous and
early sequential bilinguals are mostly manifested in the reliance on semantic
factors, which also guide early acquisition and are universal (e.g. the
Vendlerian classes). Attrition on the other hand is often described as
simplification, and is manifested in the use of the aspectual system by a
preference for less complex forms, regardless of the aspect they convey
(Armon-Lotem et al. 2006; Gagarina et al. 2005). Nevertheless, it was found
that there was a decline in the number of mistakes for both groups of
bilinguals. These findings correspond to the cross linguistic findings that
report a delay in bilingual acquisition, but not an acquisition failure (Meisel
2007, Schwartz 2004).
The development of L1 Russian tense-aspect morphology 87
1%
1%
8%
30% 11%
54%
15%
80%
Present Imperfective past Perfective past Other Present Imperfective past Perfective past Other
2. Predictions
a) Since Hebrew and Russian use the same anchor tense (Berman &
Slobin 1995) in the elicited narratives, we predict no difference in
the use of anchor tense between Russian monolinguals and Russian-
Hebrew bilinguals, despite Anstatt’s (2005) findings.
b) The use of the tense / aspect morphology in narratives (both correct
and erroneous) will be governed by a semantic bias in line with
Pereltsvaig’s (2005), Polinsky’s (1996, 1997, in press) and Anstatt’s
(2005) findings. Bilinguals will tend to assign perfective
morphology to telic verbal predicates and imperfective morphology
to atelic verbal predicates.
c) Russian-Hebrew bilinguals will experience problems with tense and
aspect morphology in the narratives, yielding higher use of the
imperfective, as found in Armon-Lotem et al. (2006) and Gagarina
et al. ( 2005, 2007).
d) Late bilinguals will have more imperfective aspect instead of
perfective and less perfective aspect instead of imperfective,
compared with early bilinguals, as shown by Armon-Lotem et al.
(2006) and Gagarina et al.( 2005, 2007). This will show in the ratio
of errors and in the distribution of errors.
e) In the distribution of errors, late bilinguals will have more tense
errors and less aspectual errors than early bilinguals, as suggested by
Armon-Lotem et al. (2006) and Gagarina et al. (2005, 2007)
findings.
f) Older children will have fewer tense and aspect errors than younger
children, in light of Armon-Lotem et al. (2006) and Gagarina et al.
(2005, 2007) findings.
3. Method
3.1. Participants
The narratives were elicited using Mercer Mayer’s wordless picture book
‘Frog, where are you?’(1969) maintaining the same procedure used in
Berman and Slobin (1994) and Strömqvist and Verhoven (2004) for a variety
of languages. Subjects were instructed in L1 Russian to look at the twenty-
four pictures depicting the adventures a boy and his dog experience when
they try to find their runaway frog. No discussions about the plot were held
with the child prior to the narration in order to avoid any influence on a
child’s narrative. The children's questions about the characters or objects in
the pictures were answered by single words such as olen’ ‘a deer’ or ulej ‘a
beehive’, while going over the pictures. Afterwards the participants were
encouraged to tell the best stories they could. The interviewer did not ask a
child any questions during the interview and provided the minimal verbal
feedback, prompting a child by nodding, ‘uh’, da ‘yes’, tak ‘so’ and
prodolzhaj ‘go on’ phrases when necessary. The participants were
interviewed individually and digitally recorded. All narrations were fully
transcribed and segmented at the level of the utterance.
The length of narrations (number of words and utterances), percentage
of Verbal Utterances (VU), i.e., utterances containing a verb, and Mean
Length of Utterances (MLU) were measured for the purposes of the study. In
addition, in order to investigate the development of L1 Russian tense and
aspect, all verbal predicates were analyzed for finiteness, tense, use of the
90 Olga Gupol, Susan Rothstein & Sharon Armon-Lotem
4. Findings
Our findings show that there is no fundamental difference between early and
late bilinguals at the two age levels in the overall organization of narratives
with respect to the MLU, percentage of VUs, anchor tense and use of
grammatical aspect. The analysis of the general characteristics of the
narratives shows that though the 7s have a somewhat larger MLU than the 4s,
there is no significant difference in the length of the narratives among
participants of the present study with respect to their age and length of L2
Hebrew exposure (see table 2). The number of verbal utterances is already
adult like at the age of 4-5, and does not increase significantly over time.
Four children (3 early bilinguals and 1 late bilingual aged 4-5) were excluded
from the study at this point since they had less than 100 words, a MLU lower
than 2 and less than 5 VUs in their narratives. The Russian-Hebrew
bilinguals’ MLU is comparable with Russian-English bilinguals’ MLU (5.1)
but is lower than the children's MLU in Standard Russian (8.5) as reported in
Polinsky for one monolingual child aged 8 (2007).
1% 12%
3%
12%
31% 31%
54% 56%
Present Imperfective past Perfective past Other Present Imperfective past Perfective past Other
C: Early 7s D: Late 7s
1%
21%
1%
17%
56%
54%
22%
28%
Present Imperfective past Perfective past Other Present Imperfective past Perfective past Other
100%
80% STATES
60% ACTIVITIES
ACCOMPLISHMENTS
40% SEMELFACTIVES
20% ACHIEVEMENTS
0%
early 4yrs late 4yrs early7yrs late7yrs
40 40
34
35 32,5
35
30 30
25 25
C: Early 7s D: Late 7s
38,4
40
40
35 31.1 35
30 30
25 25
20 17.1
20
12.6 13
15 10.9 10.6 12,8
11,9 12,7
15
10
5.5
5.5 6.1
10
4,8
5 0.1 0.1 0.1 00.2 3
5 0,4 0,3 1 0,2
0
0
PAST FUTURE PAST PRESENT FUTURE
PAST FUTURE PAST PRESENT FUTURE
PERFECTIVE IMPERFECTIVE
PERFECTIVE IMPERFECTIVE
Both early and late 4s and 7s show similar distribution of lexical classes
of states and activities in the present and past tenses in the narratives. All
groups prefer to use more states in the imperfective past (12-17%) than in the
imperfective present (5-8%). Early and late 4s use more activity verbal
predicates in the imperfective past (10% and 13%, respectively) than in the
imperfective present (5% and 6%, respectively). Early and late 7s equally use
activity verbal predicates in the imperfective past (11% and 13%,
respectively) and imperfective present (11% and 13%, respectively). The
activity predicates are generally not used in the perfective past in the
bilingual narratives. The early and late bilinguals use them in no more than in
5% of cases. Similarly, accomplishment, achievement and semelfactive
verbal predicates are rarely denoted by imperfective morphology (round 1%).
Despite the fact that the overall temporality make up is similar in the
narratives of the four age groups, there are differences in the use of the tense
and aspect morphology among these groups. Our findings show that Russian-
Hebrew bilinguals’ verb acquisition is characterized by a high rate of errors
among early and late bilinguals in both age groups. In order to investigate the
use of tense and imperfective / perfective grammatical aspect within the verb
development in general, we analyzed all verbal tokens. Table 3 shows a
significant difference (p<0.05 on a 1-tailed independent t-test) in the total
number of errors, and in the ratio of errors out of verb tokens between the 7s
and the 4s, but not within each age group. There is no significant difference
between the early and late 4-year-olds (38% vs. 26%) and early and late 7-
year-olds (20% vs. 16%). However, 4-year-old bilinguals produced
significantly more mistakes as compared with 7s. The results show an
improvement with age for early and late bilinguals.
morphology, reflexivity marking, wrong pattern formation for the stem shift,
tense, im / perfective aspect and derivational morphology. Table 4 shows that
tense and aspect mistakes prevail in bilingual narratives for both early and
late 4s and 7s. Table 4 demonstrates a significant difference (p<0.05 on a 1-
tailed independent t-test) in the percentage of the tense and aspect errors out
of all verb tokens both between the 7s and the 4s and within early and late 4s,
but not within early and late 7s. The difference in the distribution of mistakes
disappears between the early and late bilinguals with the age.
Table 4. Distribution of errors in L1 Russian elicited narrative.
* *
*
As can be seen in table 4, all children know how to assign the tense and
aspect morphology to the verbal predicates (though the early 4s had very few
Root Infinitives), but this is done erroneously in 21% (for early 4s), 17% (for
late 4s), 10% (for early 7s) and 9% ( for late 7s) out of all verb tokens. Both
early and late 7s, as well as the late 4s show the similar distribution of errors.
All three groups have significantly more errors in which imperfective aspect
is used instead of perfective (9% for late 4s, 6% for early 7s and 5% for late
7s), and a few errors in which perfective aspect is used instead of
imperfective aspect (2% for late 4s, 1% for early 7s and 1% for late 7s).
In comparison with the other groups, the early 4s have significantly
more errors in which perfective aspect is used instead of imperfective (5%).
96 Olga Gupol, Susan Rothstein & Sharon Armon-Lotem
In addition, early 4s have significantly more tense errors than early 7s (8%
vs. 3%). In general, the erroneous grammatical tense was the second major
source of mistakes for all the groups. For the purpose of the current study we
assumed that tense errors, unlike tense shifting, involve ungrammatical use of
tenses like in example 8:
(8) Kada noch’ stala, oni zhdut mamu
When the night came, they are waiting for mom. (Leya, 5; 3)
The example shows that the bilingual child cannot locate two different events
with respect to each other within one utterance. The child uses perfective past
with the achievement verb stalaPRF / PAST ‘became’, used instead of the verb
nastupilaPRF / PAST ‘came’, in the first clause of the complex utterance and the
imperfective present tense with an activity verb zhdutIMPF / PRES ‘are waiting’
in the second clause of the utterance. The inconsistency in the use of tenses
causes the non-cohesiveness of the bilinguals’ narrative, because the
temporal correlation between events is broken. The analysis of the tense-
aspect errors out of all verb errors presented a number of significant findings,
as seen in table 5:
Table 5. Distribution of errors out of other errors
Early Late Early Late
Types of the verbal errors
4s 4s 7s 7s
1. Root Infinitives (RI) 1% 0% 0% 0%
2. Error in number 11% 9% 17% 15%
3. Error in gender 1% 3% 5% 3%
4. Error in person 21% 22% 13% 20%
5. Wrong verbal ending 0% 1% 2% 2%
6. Wrong tense 13% 6% 4% 5%
7. Omission of the reflexivity marker 22% 36% 30% 31%
8. Redundant reflexivity marker 11% 10% 12% 6%
9. Wrong perfective aspect 12% 4% 7% 9%
10. Wrong imperfective aspect 9% 9% 9% 8%
Total number of errors 152 137 129 93
and 56 % for late 7s). The ungrammatical use of tense occurred in 22 % for
late 4s, 13% for early 7s and 20% for late 7s. All the groups did not differ
significantly with respect to assigning of the inappropriate grammatical tense.
Nevertheless, the distribution of aspect errors shows us a different
developmental trend in the age groups. Late 4s as well as early 7s and late 7s
show a similar distribution of aspect errors. The use of imperfectives instead
of perfectives surpasses all other mistakes (36 % for the late 4s, 30% for the
early 7s and 31% for the late 7s). Notably, these groups do not misuse
perfectives instead of imperfectives. The percentage of this kind of errors is
significantly lower as compared with the use of the imperfective instead of
the perfective morphology (6 % vs. 36 % for the late 4s, 4% vs. 30% for the
early 7s and 5% vs. 31% for the late 7s).
The early 4s show a different use of aspect morphology as compared
with the other groups. They use inappropriately the perfective aspect in 13%
of cases when the imperfective aspect should be used. This result is
significantly higher than that among the late 4s (6%), the early 7s (4%) and
the late 7s (5%). Unlike the other three groups, early 4s have significantly
less errors in which imperfective aspect is used instead of perfective aspect
(22% as compared with 36 % for the late 4s, 30% for the early 7s and 31%
for the late 7s). Nevertheless, the tendency in the correlation of aspect errors
remains the same for early 4s as for the other three groups. They tend to
assign imperfective instead of perfective morphology in higher number of
uses (22%) than perfective instead of imperfective morphology (13%) to the
verbal predicates. The erroneous use of grammatical aspect is influenced by
the telic / atelic semantic properties of the verbal predicates (see figure 10).
Figure 10. Distribution of the erroneous use of the( im)perfective grammatical aspect
by lexical classes of verbs
100%
80% STATES
ACTIVITIES
60%
ACCOMPLISHMENTS
40% SEMELFACTIVES
20% ACHIEVEMENTS
0%
Early 4s Late 4s Early7s Late7s Early 4s Late 4s Early7s Late7s
7s, as well as late 4s, showed the same distribution of errors. All three groups
had significantly more errors in which imperfective aspect was used instead
of perfective and a few errors in which perfective aspect was used instead of
imperfective aspect. In comparison with the other groups, early 4s had
significantly more errors in which perfective aspect was used instead of
imperfective, and significantly less errors in which imperfective aspect was
used instead of perfective aspect. Our findings showed that the erroneous use
of imperfective and perfective morphology was not accidental and was
strongly governed by a semantic bias.
References
M. Rafael SALABERRY
University of Texas-Austin, USA
1. Introduction
In Spanish, both single and iterated telic events can be conveyed with the use
of the Preterite (1a and 1b), whereas habituality is typically expressed with
the use of the Imperfect (1c) 1. As the translated versions of these sentences
show, English does convey the contrastive meanings of iterativity and
habituality, although the formal means to express those meanings are
different. In English, iterativity and habituality may be contrasted with the
use of the Simple Past versus the periphrastic as (1b) and (1c) show. Notice,
however, that (1c) shows that habituality may be expressed with the Simple
Past as well, thus generating a possible source of confusion that is not the
case in Spanish.
(1) a. Ayer el tren del mediodía llegó (PRET) tarde.
Yesterday the 12 o’clock train arrived late.
b. Durante muchos meses el tren del mediodía llegó (PRET) tarde.
For months the 12 o’clock train arrived late.
c. Cuando era niño el tren del mediodía llegaba (IMP) tarde.
When I was a child, the 12 o’clock train used to arrive / would arrive /
arrived late.
1
Sample sentences adapted from examples used in Slabakova and Montrul
(2007).
2. Aspect
Aspect “concerns the different perspectives which a speaker can take and
express with regard to the temporal course of some event, action, process,
etc. ” (Klein, 1994: 16). This definition captures what is most essential about
a definition of aspect: the notion of particular interpretations of temporality
of situations in reality conveyed by or depicted in linguistic terms. Klein’s
definition is, however, overly broad and requires further refinement. For
instance, most definitions of aspect (including Klein’s extended discussion of
the topic) postulate a division of aspectual phenomena into two levels that we
can define as (inherent) lexical aspect (ontological distinctions or situation
types) and grammatical aspect (viewpoint or speaker’s perspective) 2. Smith
(1997: 5), for instance, argues that “[s]entences present aspectual information
about situation type and viewpoint. Although they co-occur, the two types of
information are independent.” For the purpose of this paper, I will
concentrate on the analysis of viewpoint aspect (using Smith’s terminology)
or grammatical aspect. The distinction about situation aspect (or lexical
aspect in other publications) and grammatical aspect is, however, an
important one that I will address in more detail in the discussion section in
the context of the analysis of proposed theoretical arguments about the
development of knowledge about iterativity.
Both English and Spanish convey information about the contrastive
values of aspectual meanings explicitly through morphological means. For
instance, in Spanish, the Preterite-Imperfect morphological contrast serves to
communicate various aspectual meanings among which the concept of
boundedness is perhaps the most basic one. In order to depict an eventuality
as bounded, we need to make reference to the boundaries of an eventuality
(Depraetre, 1995). Thus, in sentence (2a) the Preterite is used to depict the
eventuality of eating an apple with a terminal point that has been achieved,
whereas (2b) uses the Imperfect to depict the same eventuality without the
terminal point.
(2) a. Julián COMIÓ una manzana.
b. Julián COMÍA una manzana, cuando LLEGÓ Lucas.
2
The label lexical aspect is a misnomer of sorts, if its meaning is restricted
simply to the aspectual information conveyed by the lexical predicate (in the
most basic interpretation of Vendler’s classification of verb types). Given that
more than just predicate level information is necessary to identify the basic
aspectual nature of a situation or eventuality (e.g., Verkuyl, 1993), the term
situation aspect proposed by Smith (1991, 1997) may be a more accurate label.
Along the same lines, Filip (1999: 15) prefers the term “eventuality type instead
of aspect, aspectual class or inherent lexical aspect to minimize confusion with
the other category.”
Theoretical views on the development of Spanish L2 knowledge 109
punctuality ⇓
continuousness ------------------⇒
punctuality ⇓
habituality ⇒⇒⇒⇒
iterativity ⇓⇓⇓⇓
vertical arrows in the graphic serve to convey the notion that the concept of
iterativity maintains the focus on the terminal points of the various iterations
of the event. For instance, Langacker (1999) proposes that iterativity
(repetitive in his terminology) is different from habituality to the extent that
iterated eventualities are anchored to specific points in time, whereas habitual
eventualities are not (for an extended analysis see Comrie, 1986; Langacker,
1987, 1999; Salaberry 2008).
Iterativity and habituality tend to be confused as one and the same
concept (in some cases used as synonymous expressions) in part because they
represent two specific nuances of meaning of the basic concept of the
repetition of eventualities. Several researchers, however, have pointed out
some distinctions between the two concepts (Comrie, 1986; Depraetre, 1995;
Langacker, 1987, 1999). Depraetre (1995: 12), for instance, points out that a
habit or repetitive situation is not inherently atelic and unbounded. In fact, a
repetitive situation can also be telic, the situation holds a predetermined
number of times. The following examples from Smith (1997: 51) are useful
to illustrate the telic meaning of various iterated eventualities:
(4) a. Last year, the train arrived late.
b. For months, the train arrived late.
(5) a. Last year, John fed the cat.
b. For months, John fed the cat.
(6) a. Last year, John moved to a new apartment.
b. For months, John moved to a new apartment.
The examples from Smith described in the previous section raise important
questions regarding the representational nature of habituality versus
iterativity. For instance, should the representation of both habituality and
iterativity be regarded as part of grammatical knowledge (e.g., Salaberry,
2008) ? Or alternatively, should only habituality be regarded as part of
grammatical knowledge, whereas iterativity is conceptualized as pragmatic
knowledge (e.g., Slabakova & Montrul, 2002, 2007).
According to Slabakova and Montrul, the iteration of a telic event with the
use of adverbials represents a case of a shift in meaning at the level of
pragmatics (as opposed to grammar). For instance, as an example of a
pragmatically-induced shift, Slabakova and Montrul argue that option (7b)
represents a case of “implicit coercion” of the lexical aspectual class of an
achievement verb (i.e., llegar) into a habitual activity, prompted by the use of
the adverbial clause that precedes it (e.g., durante muchos meses).
(7) a. Ayer el tren del mediodía llegó (PRET) tarde.
Yesterday the 12 o’clock train arrived late.
b. Durante muchos meses el tren del mediodía llegó (PRET) tarde.
For months the 12 o’clock train arrived late.
On the other hand, Ziegeler (2007: 999), for instance, points out that the
solution of using syntactic means to resolve semantic conflicts (i.e., coercion)
is not adequate and argues instead for a process of grammaticalization and
historical change. For instance, she contends that the generalization of the use
of the English progressive with dynamic verb types (e.g., she was winning, he
was dying) can be regarded ‘as part of the shift towards the verb end of the
noun-to-verb continuum’ in the transition of English (p. 1020). In fact, even
if some of the information about aspectual knowledge were to be regarded as
‘pragmatic’ in nature (i.e., conventional meanings in Smith’s terms), it could
eventually become part of the grammatical system. On this point, Ziegeler
surmises that “transitory pragmatic inferencing strategies may eventually be
adopted and generalised across entire populations of users, and across entire
paradigmatic domains of usage, where they may become conventionalised in
usage, and henceforth no longer identifiable as changes” (p. 1022).
In this regard, a potential expansion of Ziegeler’s proposed process of
grammaticalization of meanings can be instantiated in particular uses of the
Spanish Preterite and Imperfect in association with the notions of iterativity
and habituality. For instance, let us review the distinct representation of these
two aspectual notions in sentences (9) and (10) 3.
(9) a. Durante mi infancia, el tren del mediodía llegaba (IMP) tarde.
During my childhood, the 12 o’clock train arrived late.
b. ?Durante mi infancia, el tren del mediodía llegó (PRET) tarde
During my childhood, the 12 o’clock train arrived late.
(10) a. ?Durante muchos años, el tren del mediodía llegaba (IMP) tarde
For years, the 12 o’clock train arrived late.
b. Durante muchos años, el tren del mediodía llegó (PRET) tarde
For years, the 12 o’clock train arrived late.
4
It should be noted that there have been additional interpretations of the
distinction between habituality and iterativity as a grammatical process. For
instance, using Klein’s distinction between Topic Time (TT) and Situation Time
(TSit), Starren (2001: 150-154) equates habituals with a single TSit associated
with several TTs, whereas iteratives are represented by several TSits associated
with a single TT. More specifically, Starren argues that in sentences like [Each
winter I get a “big” anti-influenza vaccination] and [I am always -each
morning- awake at eight o’clock], “the situation time … is hooked up to a series
of topic times (quantified by the adverb = all time spans), by which the
utterance acquires a typical imperfective meaning of habituality” (p. 150). In
contrast, Starren argues, in iteratives as in [Last year I had accidents, again and
again], the utterance structure makes reference to a series of situations (complex
Tsit) “(consisting of separate phases of the situation, quantified by) linked to
one topic time” (p. 152). I believe, however, that this explanation is inadequate
to take into account examples represented by iterative sentences reviewed above
such as [For years, the train arrived late]. In essence the above-mentioned
iterative sentences have more than one TT, a fact that stands in contrast with
Starren’s explanation and that needs to be further analyzed.
114 M. Rafael Salaberry
Several researchers have proposed that the perfective form is the more basic
one of the perfective-imperfective contrast (e.g., Binnick, 1991; Comrie,
1985; Fleischman, 1990). Applied to Spanish, this claim entails that the
Preterite is the basic past tense marker, whereas the Imperfect is a more
straightforward aspectual marker. Indeed, Doiz-Bienzobas (1995, 2002)
claims that several depictions of the perfective-imperfective contrast seem to
indicate that the perfective acts like a straight tense marker, whereas the
imperfective functions as a true aspectual marker because the “Imperfect
does not interact with the boundedness of the situations” (p. 32)5. Thus, in the
following sentence she argues that the contrast in meaning between the
Preterite and the Imperfect is related to a view of the given situation from a
past viewpoint (11a) marked with the Imperfect, or alternatively from speech
time (11b), the latter marked with the Preterite.
(11) a. El sermón me parecía (IMP) eterno.
The sermon seemed interminable.
b. El sermón me pareció (PRET) eterno.
The sermon seemed interminable.
5
In keeping with Depraetre’s (1995) principled distinction between boundedness
and telicity, Doiz-Bienzobas reference to boundedness may be reinterpreted as a
reference to telicity.
6
The notion of a default past tense marker is not unique to Langacker or Doiz-
Bienzobas. Giorgi and Pianesi (1997: 174), for instance, propose that the
Theoretical views on the development of Spanish L2 knowledge 115
First, the fact that the learner is considering two different past tense forms of
the verb is a good indication that she understands the need to mark past tense
on the given verb (and also marked morphologically in English, the learner’s
native language: L1 transfer). Second, note that the learner’s repetition of
imperfective eran signals that the learner understood the form used by the
interlocutor. Despite such repetition of the imperfective form, however, the
learner insisted on using a perfective form in her answer (fueron). Finally,
notice that the imperfective form of the verb (to be) is usually the most
common in most input environments provided to learners. In other words,
frequency in the input would have favored the use of the Imperfect.
Nevertheless, the use of the perfective form was clearly preferred to mark the
verb in past tense.
To the best of my knowledge, the first study to advance the notion of a
default marker of past tense, was the one carried out by Wiberg (1996) in
which she argued for the effect of an unmarked past tense hypothesis on the
basis of the analysis of L2 Italian data. Salaberry (1997, 1999) further
ascertained the potential role of a default past tense marker among fifteen
college-level L2 Spanish learners (from second to fifth semester of
instruction) through oral retellings of a short film excerpt. The most
important finding of this study was that the least proficient learners did not
use the Imperfect during the first or the second session of data collection.
Interestingly, these learners had received explicit instruction and practice on
the use of the Imperfect during the two weeks prior to the session when they
produced their second oral narratives (the Preterite had been introduced
before the first session). On the other hand, the more experienced learners
(from third to fifth semester) used the Preterite and Imperfect in close
association with the value of lexical aspectual classes. That is, the use of the
perfective form was associated with telic events and, vice-versa, the use of
the imperfective form was associated with states. Furthermore, the influence
of lexical aspect increased constantly as a function of proficiency in the L2.
In sum, Salaberry proposed that during the initial stages of L2 development,
L1 English speakers use the Spanish Preterite as a default marker of past
tense across lexical aspectual classes. In contrast, the role of lexical aspect
becomes relevant as experience with the language increases (i.e., third to fifth
semester).
On the other hand, some studies have provided conflictive findings that
raise questions about the use of the perfective form as a default past tense
marker. For instance, Salaberry (2003) showed that in certain narratives
containing a preponderance of verbs in background clauses, the preferred
past tense form might be the Imperfect (and not the Preterite). Salaberry
attributed this finding to the confounding effect of factors such as grounding
(background) or, perhaps simply the priming effect brought about by the
preponderant use of the Imperfect in background clauses. Following up on
these conflictive findings, Salaberry (2011) compared the relative effect of
inherent lexical aspect and discursive grounding on the use of L2 Spanish
Preterit and Imperfect using a written 40-item text-based multiple-choice task
(a personalized fictional narrative). The findings of the study revealed that
whereas both lexical aspect and grounding were directly correlated with the
choice of past tense marker across all levels of proficiency in Spanish, the
effect of grounding was the only one that reached statistical significance.
Thus, Figure 3 below shows that while the proportional use of the Preterite
remains unchanged in association with foregrounded eventualities (i.e., a
default past tense), the use of the Imperfect gradually increases in association
with the backgrounded eventualities in the narrative.
Theoretical views on the development of Spanish L2 knowledge 117
0.9
0.8
0.7
0.6
0.5
foreground
background
0.4
0.3
0.2
0.1
0
2nd sem. 3rd sem. 4th sem. 5th sem. Natives
Although the possible effect of a default marker of past tense has been
proposed to account for empirical data from beginning learners (e.g., Rocca,
2005; Salaberry, 1999, 2003; Wiberg, 1996), more recent studies have
provided evidence to extend its application to more advanced levels of
proficiency (e.g., Antonio, 2007; Granda, 2004; Liskin-Gasparro, 2000;
Lubbers-Quesada, 1999, 2007) 7. In this respect, the concept of iterativity
(representing a more complex learning target than simply punctuality or
habituality) becomes a viable grammatical target to ascertain the possible
effect of a default past tense marker throughout development.
Among one of the first studies to assess the nature of the acquisition of
knowledge about iterativity and habituality, Slabakova and Montrul (2002,
7
Arguably, the Default Past Tense Hyothesis may also be relevant for the
analysis of data from natural learners (e.g., Giacalone-Ramat, 2002) and
heritage learners (e.g., Silva-Corvalán, 1994).
118 M. Rafael Salaberry
2007) analyzed data from 60 English native speakers (27 advanced and 33
intermediate) and 27 native Spanish speakers. The test instrument was a
grammaticality judgment test that consisted of 49 pairs of sentences testing
six conditions of shifted aspectual interpretations and one set of seven
distractor sentences. The sentences were presented in pairs to test the
meaning contrast of verbs with a basic and a shifted interpretation brought
about by grammatical or pragmatic means. For the purpose of our discussion,
we will only analyze the findings related to the outcome with achievements
in their default punctual representation (sentence 7a above) and their
“shifted” interpretation into iteratives with the addition of an adverbial
(sentence 7b above). The results of the study (summarized on Table 1)
revealed that both advanced and intermediate learners as well as the native
speakers consistently accepted the use of the Preterite with punctual
meanings of the verb (the single arrival of the train). The scores for
intermediate and advanced learners and natives were 4.33, 4.88 and 4.88
respectively. That is, measured on the 5-point scale, all groups strongly
agreed with the grammaticality of that sentence.
Pérez-Leroux et al. analyzed data from 31 students and ten native speakers on
the effects of unique, habitual and iterated situations. The learners were
divided into two levels of proficiency based on course enrollments in second
or third year (13 and 18 respectively). The test instrument was a 30-item
grammaticality judgment task. Each one of the four target conditions was
represented by five examples making up a total of 20 target sentences (there
were also 10 distractor sentences). As shown above, the conditions that
triggered an iterative interpretation were: explicit with the use of an iteration
adverbial (e.g., repetidamente) and implicit with the use of a duration
adverbial (e.g., por días). The following are samples of the test sentences
used by Pérez-Leroux et al. in their study (# = unlikely, less acceptable, √ =
preferred):
(15) a. En su niñez, jugaban (√ IMP) en la calle por las tardes.
b. En su niñez, jugaron (#PRET) en la calle por las tardes.
In their childhood, they used to play / played in the streets in the afternoon.
(16) a. El terremoto sacudió (√PRET) la ciudad a las 8.
b. El terremoto sacudía (#IMP) la ciudad a las 8.
The earthquake shook the city at eight.
(17) a. El terremoto sacudió (√PRET) la ciudad por días.
b. El terremoto sacudía (# IMP) la ciudad por días.
[iterative, implicit coercion]
The earthquake shook the city for days.
120 M. Rafael Salaberry
Both the findings from Slabakova and Montrul and Pérez-Leroux et al.
reviewed above seem to indicate that L2 learners of Spanish are not capable
of distinguishing the contrast of iterativity versus habituality expressed in
Spanish through the use of Preterite and Imperfect respectively.
8
Even though Pérez-Leroux et al. argue that there was a gradience in favor of the
unique events (“advanced speakers were able to clearly reject the use of
Imperfect in unique contexts”) the statistical tests did not support this
contention.
Theoretical views on the development of Spanish L2 knowledge 121
On the other hand, the findings from Slabakova and Montrul and Pérez-
Leroux et al. need to be put in perspective given that the learners who
participated in these studies may have not been advanced enough to
demonstrate knowledge of iterativity as opposed to habituality. For that
purpose, Salaberry and Scholes (submitted) analyzed data collected among
more advanced learners (graduate students teaching Spanish) as well as
students whose levels of proficiency are approximately equivalent to the ones
used in previous studies (second and third year academic courses). Salaberry
and Scholes assessed knowledge of iterativity versus punctuality with 24
prompt sentences used by Slabakova and Montrul in their study 9. Salaberry
and Scholes specifically asked participants to rate the grammaticality of each
one of the target sentences on a scale from 1 (completely ungrammatical) to 4
(perfectly grammatical) 10. The mean scores for each group according to the
condition represented in the study are shown on Table 2 below.
The findings from the study show that no learner group, including the most
advanced graduate students demonstrated knowledge of the concept of
iterativity as distinct from habituality. An outstanding question, however, is
whether non-native speakers made progress in the more basic concept of
boundedness, also conveyed through the use of Preterite and Imperfect. Thus,
in addition to the test of iterativity described above, Salaberry and Scholes
used a 30-item test based on the straight selection of Preterite or Imperfect in
the context of a typical classroom test (test based on Psycho movie used in
9
This procedure ensured a high degree of confidence that the results between
studies were roughly comparable. Nevertheless, Salaberry and Scholes added a
second component to their study: all participants were asked to provide a brief
explanation of the rationale for their judgment.
10
Note that Slabakova and Montrul used a 5-point scale. The 4-point scale was
deemed to be more appropriate to lead participants to make a categorical
decision (i.e., acceptable versus non acceptable). Given the focus of the study on
the more advanced learners, the lack of availability of a “not sure” option was
not regarded as crucial. More importantly, the fact that informants were asked to
describe what they thought it was not grammatically appropriate represented an
adequate measure to determine if learners were indeed focusing on the target
notion assessed in the study (iterativity versus habituality).
122 M. Rafael Salaberry
What is important to notice in Table 3 is that the results of the traditional test
revealed stratified levels of ability among nonnative speakers according to
proficiency, with the highest level (i.e. graduate students who were teaching
Spanish) approximating native speaker performance 11. In contrast, the results
from the test of iterativity in Table 2 show no such linear development
according to proficiency level. In sum, there is a disconnection between the
results of the traditional test and the iterativity test. The lack of progress in
the understanding of iterativity versus habituality stands in contrast with their
overall linear progress towards understanding the overall use of Preterite and
Imperfect, in the context of other more prototypical uses.
6. Discussion
The empirical findings reviewed so far are based on a limited set of data, thus
they do not lead us to making any categorical conclusions. On the other hand,
the results are enticing enough to support various theoretical proposals that
are worth discussing in more detail. Overall, the combined results of the three
studies on the learning of iterativity versus habituality, reviewed above,
indicate that learners do not achieve native-like knowledge about the
representation of iterativity versus habituality in Spanish past tense
11
These data confirm the existence of at least a minimal degree of separation
between the levels and therefore substantiate the classification system used to
determine proficiency groupings.
Theoretical views on the development of Spanish L2 knowledge 123
12
Note that the previous literature on the role played by adverbials in the
composition of aspectual information has been vague. For instance, Smith
(1997: 2, emphasis added) argues the following: [s]ituation type is conveyed by
the verb constellation, which I define as a main verb and its arguments,
including subject. Viewpoint is conveyed by a grammatical morpheme, usually
verbal. Adverbials may give relevant information…
124 M. Rafael Salaberry
13
Alternatively, within the framework of Labeau’s (2005) analysis, we should
expect a movement back and forth (a “pendulum swing”) of representational
values assigned to specific aspectual markers.
126 M. Rafael Salaberry
7. Conclusion
The outcome of all three studies reviewed above seems to indicate that L2
learners of Spanish are not capable of distinguishing the contrast of iterativity
versus habituality expressed in Spanish through the use of Preterite and
Imperfect respectively. Among the major theoretical claims proposed to
account for this empirical outcome so far, the proposal about the dissociation
between grammatical and pragmatic means was discounted due to the highly
constrained interpretation of the semantic meaning assigned to aspectual
contrasts. On the other hand, I have favored the claim about the possible
effect of a default past tense marker as the empirical reflex of an
impoverished representational system of aspectual distinctions in L2 Spanish.
In this respect, Salaberry and Scholes’ proposal is very similar to the one
proposed by Pérez-Leroux et al, except for the fact that Salaberry and
Scholes assume that the default marker of past tense is relevant for both
beginning as well as experienced users of the L2 (see also Salaberry 2005,
2008). That is, the apparent role of a default marker of past tense among
proficient speakers of the L2 may be indicative of an L2 system that is
differentially represented among L1 and L2 speakers.
Theoretical views on the development of Spanish L2 knowledge 127
References
Kathleen BARDOVI-HARLIG
Indiana University
1. Introduction
(Andersen & Shirai 1996: 558). Each describes a different potential route in
the development of the progressive in second language acquisition. The
comparison of the two potential paths suggest these research questions:
Beginning with the prototypical association of progressive with activities,
does progressive retain its core meaning of process and begin to associate
with other aspectual categories ? Or starting at the same point, does the
progressive begin to show iterative, habitual, and futurate readings while still
associating with activities ? In other words, is the expansion of the meaning
of the progressive in second language acquisition better characterized by
movement across lexical aspectual categories or by the integration of
iterative, habitual, or futurate meanings within categories ?
The following sections present a brief overview of the meanings
attributed to the progressive in English, review the L2 research on the
acquisition of the progressive in English and other languages, and then report
on the expansion of the progressive from prototypical to nonprototypical uses
in a longitudinal study of 16 learners of English as a second language.
Binnick (1991: 281) observes that “the English progressive has proven
intractable and its analysis controversial...no one has ever specified in a
completely satisfactorily general way how it is used,” adding that “no one has
convincingly argued for any one basic meaning for it, but neither has anyone
established that it lacks one” (pp. 281-282) 1. As an illustration, consider
Mindt’s (2000) account of the progressive which identifies no fewer than
nine meanings, including (1) Incompletion, (2) Temporariness, (3) Iterative /
habit, (4) Highlighting / prominence, (5) Emotion, (6) Politeness / down-
toning, (7) Prediction [future], (8) Volition / intention [future], and (9)
Matter-of-course. More conservatively, Quirk, Greenbaum, Leech, and
Svartvik (1985: 198) list three main components of the meaning of the
progressive, not all of which need to be present in a single reading: (1) the
happening has duration, (2) the happening has limited duration, and (3) the
happening is not necessarily complete.
Binnick (1991) identifies four major characterizations of the
progressive: durative (grammatical) aspect; action in process or progress;
incompletion (or lack of necessary completion); and progressive of the frame
(providing a frame for another event or situation). Binnick provides
counterexamples to each one, supporting the claim that no one approach
adequately accounts for all meanings of the progressive.
1
It is certainly fortunate for second language learners that acquisition does not
depend on a satisfactory linguistic analysis of a given target.
A longitudinal investigation of the progressive prototype in L2 English 133
2
In contrast, Dowty (1979: 157) does attempt to provide a formal account of both
the progressive and futurate. He proposes to “treat the ‘futurate progressive’ as
imperfective progressive combined in a purely compositional way with a
sentence in the tenseless future.” Such an account is beyond the scope of the
current paper.
3
I recognize the potential mismatch between British corpora and the analysis of
production data of learners of American English. There is no comparable corpus
analysis of American English available for the progressive; Römer’s work
(Römer, 2005; Nusselhauf & Römer, 2007) focuses solely on the progressive
and has linguistic detail that is compatible with linguistic analysis of
interlanguage.)
134 Kathleen Bardovi-Harlig
Predicate type plays a significant role in the use and meaning of the
progressive. The progressive is unremarkably associated with activities
(ACT) and accomplishments (ACC). In fact, this association is made use of
in tests of lexical aspectual categories as in the progressive test described by
Dowty (1979: 55). Statives disallow progressive as shown in (3a).
(3) Only non-statives occur in the progressive:
a. *John is knowing the answer (STA)
b. John is running (ACT)
c. John is building a house (ACC)
Achievements are not subject to the test (which Dowty annotated with “does
not apply”) because the progressive with achievements forces or coerces a
repeated reading. As Binnick (1991) observes, the progressive occurs with
achievements (ACH) in series (i.e. repeated events), as in (4).
(4) a. Sue is jogging (ACT)
b. Sue is repairing the plumbing (ACC)
c. Sue is noticing a lot of drunks on the street these days (ACH series)
(6) Because he is going home all summer (Guillermo, Written, Month 8.0)
Although the context suggests that he didn’t intend it, the learner’s use of a
future progressive nevertheless triggers a repeated reading. Go home, is telic
(an accomplishment like other activities plus locatives), and so it is open to
the reading that he will go home over and over again; an alternative reading
draws on the duration of the event to suggest that “going” or traveling will
take all summer. What the learner meant was that he would be home all
summer (the result of going there), but a resultative reading is not available
from the sentence as written without generous interpretation from the reader.
Note that adverbials play an important supporting role in the interpretation of
the progressive. This is well-documented in the tense-aspect literature as well
as in the acquisitional literature (in tense-aspect theory, Dowty 1979; Smith
1997).
Given the strong association of verbal morphology and lexical aspect in
adult native-speaker linguistic systems, it is not surprising that second
language acquisition also shows an interaction of verbal morphology and
lexical aspect. This is captured in the well-known aspect hypothesis: First and
second language learners will initially be influenced by the inherent semantic
aspect of verbs or predicates in the acquisition of tense and aspect markers
associated with or / affixed to these verbs (Andersen & Shirai 1994: 133).
The aspect hypothesis can be broken down into four separate claims
regarding perfective past, general imperfective, progressive, and progressive
with states (Shirai 1991: 9-10; see also Andersen & Shirai 1996: 533). Of
particular interest to this investigation is the specific prediction concerning
the progressive. “In languages that have progressive aspect, progressive
marking begins with activities, then extends to accomplishments and
achievements.” As shown in the section on second language acquisition, the
aspect hypothesis has been supported in L2 acquisition, but the use of the
progressive with achievements is rare, so the later stages of the hypothesized
developmental path have not been documented in detail.
This study compares the two prototype accounts of the progressive, from the
perspective of the association of grammatical aspect with lexical aspectual
category (the aspect hypothesis) and the meanings of the progressive itself, as
potential accounts of development as seen in a corpus of learner data from a
longitudinal study. Before considering the longitudinal data, the next section
briefly reviews studies of the progressive in L2 acquisition.
4
There are cases that are ambiguous between future and present readings, so the
intended reference to future might be somewhat higher, but cannot be discerned
from the data.
5
Not all linguists adopt the Vendler categories and the corresponding definitions
of and test for lexical aspectual categories. Biber and colleagues (1999) offer
examples of progressive with states, many of which would be activities by the
tests used by Dowty (1979) and others.
6
32 of the 125 activities are mental process verbs. Giacalone Ramat reports these
separately because they are a relatively large group, but states that they are
activities. Dividing the activities into two group yields 63% activities and 22%
with mental processes.
A longitudinal investigation of the progressive prototype in L2 English 137
80
70
60
% U s e o f P r o g r e s s iv e
50
ACT
40 ACC
ACH
30
20
10
0
BW FW BO FO
Grounding by Mode
7
Many of the studies reviewed here investigated at least the perfective past in
addition to the progressive.
138 Kathleen Bardovi-Harlig
dynamic verbs in the background than in the foreground, and greater for
activities than accomplishments, and rare for achievements. (Out of 2,779
predicates, 185 of which are progressive, only 15 occur with achievements.)
Using a within-category analysis reveals that in the background of written
narratives, 67% of activities occur with the progressive and in the foreground
13% occur with the progressive. In oral narratives, 41% of background
activities occur in the progressive, and 20% of the foreground activities also
do. In the written and oral narratives, respectively, 28% and 26% of the
background accomplishments occur in the progressive; but only 1% and 5%
of achievements occur in the progressive in the written and oral narratives,
respectively. This is illustrated in Figure 1 (based on data reported by
Bardovi-Harlig 1998). O stands for oral, W for written, F for foreground, and
B for background). Using an across-category analysis shows that in the
written narratives 80% of the progressives occur with activities; in oral
narratives 70% of progressives occur with activities. In other words, the
progressive occurred more often in the background than in the foreground,
and more often with activities than any other aspectual category.
Shirai's (1995) study of three Chinese learners of Japanese as a second
language enrolled in an intensive Japanese program also supports the aspect
hypothesis; these learners also showed dominant use of progressive -te i-
with activities (55% of all uses of -te i- occur with activities). In a second
study of Japanese, the results of a judgment task administered to 17 tutored
Chinese learners of Japanese suggested that "learners found it easier to
recognize the correctness of -te i- with activity verbs" than with achievements
(Shirai & Kurono 1998: 264). The interesting fact about -te i- is that it has a
progressive reading with activities and a resultative reading with
achievements. Suguya and Shirai (2007) investigated the effect of L1 on the
acquisition of the progressive with L2 learners of Japanese from progressive
and nonprogressive languages. The results of a judgment task showed that
regardless of L1, learners responded more appropriately to progressives than
resultatives and low-level groups judged progressive use more correctly for
activities than accomplishments or semelfactives. In the oral production task
progressive readings of –te i- were associated with activities and resultative
with achievements, consistent with the aspect hypothesis.
The first study to investigate the progressive using a prototype account
was Gass and Ard (1984). A series of judgment tasks (judging sentences in
isolation, sentences in short dialogues, and sets of sentences) administered to
L1 Japanese and Spanish learners of English at a range of proficiency levels
(from levels 1 to 6 of a six level intensive English program) showed that
learners rated more prototypical uses such as the action-in-progress reading
of That man is drinking cider higher than nonprototypical uses such as the
futurate John is traveling to New York tomorrow. The ungrammatical or
A longitudinal investigation of the progressive prototype in L2 English 139
anomalous readings with states such as Mary is being in Chicago now were
rejected outright when presented in isolation.
Huang (1999) investigated the progressive in the framework of the
aspect hypothesis and added to that the encoding of unitary and repeated
predicates, which adds another dimension to progressive meaning beyond the
association with aspectual categories. Five L1 Chinese learners and three
native-speakers of English participated in 60-90 minute interviews. (Neither
the interviewer nor the interview was described, although the learner and
native-speaker interviews were reported to be similar.) An across-category
analysis was conducted. Across-category analysis is sensitive to the number
of predicates produced in each category; thus the most populated category
will show the highest percentage of use (Bardovi-Harlig, 2002). Learner
production favored the association of progressive with activities to a greater
extent than native speakers did: 76% of all progressives occurred on activities
for learners and 49% for native speakers using a type count (a token count
differs only slightly with 73% and 46%, respectively). This is consistent with
previous findings. The study adds to our understanding of the acquisition of
the progressive by considering its use with unitary and repeated predicates.
Learners and native speakers show opposite distributions: 73% of the
progressives that learners produce occur with unitary predicates and 27%
occur with repeated predicates. In contrast, native speakers produce 31% of
their progressives with unitary predicates and 69% with repeated predicates.
The distribution is difficult to interpret, however, because predicates that are
traditionally considered to be activities were counted as repeated 8.
Because Huang employs an across-category analysis, it is not possible
to separate the number of repeated types or tokens from the report of verb
morphology; nevertheless, there seems to be quite different use between
native speakers and learners (but it may come from the NS using more
repeated events than learners, regardless of marking; only a within category
analysis will tell if repeated events are encoded differently from unitary
events). As Huang (1999: 124) observes, “The use of progressive
morphology in describing unitary situations focuses on the continuous,
ongoing nature of a situation.”
In a longitudinal study of future expression in L2 English, Bardovi-
Harlig (2004a, 2004b) found that the progressive is used very rarely as the
8
Huang’s classification of “repeated” seems to differ from classic aspectual
categorization in that predicates that are comprised of a verb plus an indefinite
object such as bake cookies, do homework, and make records are typically
classified as activities whereas Huang classified them as “repeated.” A more
traditional analysis would regard bake cookies as a single event unless otherwise
indicated as in She baked cookies every week in which each week’s cookie
baking would be considered a single occurrence.
140 Kathleen Bardovi-Harlig
This study investigates how the meanings associated with the progressive
spread in interlanguage from the most prototypical to less prototypical
readings, and is particularly concerned with repeated and futurate readings 10.
When presenting the progressive prototype
Process (activity > accomplishment) > iterative > habitual or futurate >
stative progressive
Andersen and Shirai (1996) note that achievements are not included in the
process category because when used with progressive they usually have an
iterative, habitual, or futurate reading11. Thus, one way for the interlanguage
grammar to expand the progressive from process to iterative or habitual
readings would be to extend the progressive to achievements as the aspect
9
The order of preference of use in future in L2 English is will (63% in the oral
corpus), then lexical futures (16%; verbs with future meaning such as hope and
plan), going to (7%), base/present (8%), and progressive (1%).
10
It was anticipated that stative-progressives, the most peripheral meaning in the
hypothesized progressive prototype would not be common in this corpus, as
predicted by hypothesis 4 of the aspect hypothesis (Progressive markings are
not incorrectly overextended to statives). This prediction seems to be supported
by learner production on cloze passages (Bardovi-Harlig & Reynolds 1995) and
narrative production (Bardovi-Harlig 1998).
11
Andersen and Shirai (1996) noted that their use of futurate included preliminary
stages of events, examples of which appear in (4). This is not the usual
interpretation of “futurate,” but because they state (p. 558, fn12) that Gass and
Ard’s hierarchy of process→ futurate→stative progressive “is in agreement
with ours”. I assume that they also include future reference as in (5) in this
category.
A longitudinal investigation of the progressive prototype in L2 English 141
4.1. Method
The data for this study were collected from 16 adult learners of English as a
second language from daily journals, compositions, oral and written
narratives, and oral interviews, yielding 1,576 written texts and 175 oral
interviews over a span of 15 months.
The learners in this study were enrolled in an intensive English program
at a university in the American Midwest. They represented four language
backgrounds (Arabic, Japanese, Korean, and Spanish). Participants were low-
level learners, as measured by their placement in the first of six instructional
levels: all but one learner completed the program at Level 5 or 6.
4.2. Analysis
The predicates in the accumulated language sample for each learner were
coded for morphology, temporal reference, and adverbials. The progressives
were then coded for aspectual repeatedness and lexical aspectual category.
All verb forms were coded morphologically. This report includes only tokens
of present progressive and past progressive. Bare progressives (V+ing with
no auxiliary) were excluded from this analysis. The second conjuncts of
conjoined predicates such as “I'm sitting here in my apartment and Ø+
thinking about the future” (Saleh, Month14.5) were not included in the totals
because the second predicate is always ambiguous between a bare
progressive and a tensed progressive since the auxiliary has been deleted.
All predicates were also coded for temporal reference (cf. Bardovi-
Harlig, 2000). Of particular relevance here are tokens of future expression.
There were 3,737 tokens of future reference encoded by various means:
2,567 in the written corpus and 1,170 in the oral corpus. The progressive
future, or futurate, was coded separately from the go-future (realized in
English as be + going to + Verb; Bardovi-Harlig, 2004a, 2004b, 2005a).
Adverbs were also coded for each predicate in the corpus. They include
adverbs of frequency, duration, time, and when clauses. An adverb-to-verb
ratio (Bardovi-Harlig 1992) was calculated for all progressives, repeated
progressives, and futurate progressives.
Repeated progressives were coded next as a single category, following
Römer (2005): these were identified in context by their adverbs or simply by
their predicates (e.g., Many audience think composers are always playing
[kidding] by music; Idechi, Written, Month 7.5; and Nobody’s waking me up
142 Kathleen Bardovi-Harlig
in the morning. I have to be ready; Eduardo, Oral, Month 5.0). Each repeated
progressive was also coded for lexical aspectual category following Dowty
(1979).
4.3. Results
There were 466 present progressives in the written corpus, and 190 in the oral
corpus, and 301 past progressives in the written corpus, and 139 in the oral
corpus, for a total of 1,096 tokens (Table 1). Out of 1, 096 tokens, 33 tokens
of progressive (3.0%) were used in repeated actions (whether iterative or
habitual) and 54 progressives (4.9%) had future reference (Table 2).
Table 1. Distribution of progressive tokens
Mode
Form Written Texts Oral Interviews
(n=1,576) (n=175)
Present Progressive 466 190
Past Progressive 301 139
Total 767 329
The 33 repeated predicates were coded for lexical aspectual category: five
telic predicates, four accomplishments and one achievement, have clear
repeated interpretations (Examples 8-11).
(8) Nobody’s waking me up in the morning. I have to be ready. [ACC]
(Eduardo, Oral, Month 5.0)
(9) Three men were once at a bridges. They were saying to everybody who
crossed it – you must tell us where are you going and you will can to pass,
but if you lie as about it, we will kill you. [ACC] (Eduardo, Written, Month
2.5)
(10) At last the bicycle toke the balles then play alon or juggle alon he was
throwing and receiving the balles by its pedals. [ACC] (Zayed, Written,
Month 10.5)
(11) If I am seeing that picture then I can relax and become calm. [ACH]
(Noriko, Written, Month 8.0)
With telic predicates, the progressive can force the repeated reading even
without an explicit adverb as in (9) and (10). It may be argued that the
indirect object to everybody in (9) makes the predicate saying to everybody
an activity because the action is distributed across many times. There was
one case in the corpus where a generic subject rendered an achievement
predicate leave home an activity as in Most Korean women, when they get
married, they are leaving their hom (Ji-An, Written, Month 4.5). In Example
(10) the paraphrase of juggle (an activity) as throwing and receiving the balls
shows that although the individual predicates throw the balls and receive the
balls are accomplishments, their use in the progressive results in a derived
repeated activity made up of telic predicates (Smith 1997). The final example
in (11) is less felicitous than the other examples because of the progressive
144 Kathleen Bardovi-Harlig
form of the achievement see. However, the repeated reading is carried by the
conditional 12.
Adverbs may also create a frame, but not occur in the same clause as the
progressive. In Example (16) always creates an adverbial frame for both was
paying attention in (16a) and was eating good food in (16b).
(16) When in my country, I was used to eating in the way how my mother cooks
that is very condimented in comparation with here. a) She was always
paying attention at our alimentation. b) She tried to be sure that our family
was eating good food. c) Right now, I have to care about what I am eating if
it is good or not by myself. (Eduardo, Written, Month 6.5)
12
Conditionals, including if-then, are classified as adverbials by Thomspon and
Longacre (1985).
A longitudinal investigation of the progressive prototype in L2 English 145
Smith observes that duration adverbials cue repeated action when the length
of time specified exceeds the length of time it usually takes to complete an
event. There were three such uses. In (18) call and two days set the frame for
I was talking; similarly in (19) with for all the afternoon and part of the night
support the repeated action of make a phone call; and in (20) with a week
sets the scene for repeated dancing, singing, and playing game.
(18) I just called my family for two days. I was talking with my brother about
that problem (Saleh, Oral, Month 12.5)
(19) Today, after to take the lunch, I came back to my room for to call to
Venezuela and I was making it for all the afternoon and part of the night but
to the end I [page] couldn't speak with my family because the phone always
was busy. (Carlos, L1 Spanish, Written, Month 1.5)
(20) And same time a festival open May, everyear. A festival continue a week.
We are dancing, singing, and playing game with together (Ji-An, Written,
Month 10).
Like the uses of when above, the conditional in (11) repeated here as (23)
assures repeated reading by establishing a habitual.
(23) If I am seeing that picture then I can relax and become calm. (Noriko,
Written, Month 8.0)
Only four tokens (12%) appeared with no adverbial support. Among them are
two examples in (9) and (10). As this section shows, the typical repeated
progressive co-occurs with an adverbial which supports or encourages the
repeated reading; although adverbs of frequency are the most common,
different types of adverbials may occur.
4.3.2. Futurate
The progressive as future is more common than the repeated readings in this
corpus; 8% (54 / 647) of the present progressives were future expressions.
But it is relatively rare in the context of future expression. Consider that of
146 Kathleen Bardovi-Harlig
The production data show that not all learners use both repeated and futurate
progressive (Table 5). Five learners show no tokens of repeated uses and four
learners show no uses of the futurate. Only one learner, Sang Wook, shows
use of neither one.
Toshihiro (L1 Japanese) and Guillermo (L1 Spanish) both show greater
than 10 more uses of the futurate than the repeated progressive, and Ji-An
(L1 Korean) shows five. Two learners show almost equal use of the repeated
and futurate progressives, Hiromi (L1 Japanese) and Eduardo (L1 Spanish).
Among learners who show fewer than five uses of either form, there are
learners whose production data show repeated progressives exclusively and
learners who show futurate progressives only. No particular pattern seems to
characterize a particular first language. If one is a prerequisite for the other, it
is not apparent in these production data, as either repeated progressive or
futurate progressive may appear as the first nonprototypical use.
13
Going and coming occur around 1,000 times per 10 millions words, whereas the
next most common set, doing, taking, and getting occur around 500 times.
A longitudinal investigation of the progressive prototype in L2 English 147
Learner L1 Progressive
Predicates Repeated Futurate
Khaled Arabic 93 1 3
Saleh Arabic 110 4 0
Abdullah Arabic 93 0 1
Hamad Arabic 23 0 2
Zayed Arabic 73 1 1
Noriko Japanese 105 4 1
Hiromi Japanese 106 7 5
Kazuhiro Japanese 30 1 1
Satoru Japanese 10 0 2
Toshihiro Japanese 58 1 10
Idechi Japanese 26 3 0
Ji-An Korean 32 2 7
Sang Wook Korean 22 0 0
Carlos Spanish 114 2 1
Guillermo Spanish 40 0 11
Eduardo Spanish 161 7 9
TOTAL 1,096 33 54
The adverb to verb ratio for repeated readings is quite high, .94, or just over 9
adverbs for every 10 verbs. Thus, when the progressive expands from the
prototypical process reading, it is not to achievements as the aspect
hypothesis predicts, but rather to repeated readings with activities through the
use of adverbials. Learners still favor the use of progressive with activities,
and the use of adverbials leads to repeated readings. The resulting multiple-
event activities constitute a derived category which encodes a series of
repetitions with an arbitrary endpoint (Smith 1997). The sub-events may be
of all event types, although the majority in this learner corpus has durative
sub-events. Crucially, as Römer (2005) explains, for native-speaker use of
progressive, continuousness and repeatedness are independent features of the
progressive; this is also the case when interlanguage allows repeated readings
for the progressive.
Whereas the role of the adverbials with telic verbs is to support the
repeated interpretation, with activities the adverbials force the repeated
interpretation. They break the one-to-one association of activities plus
progressive as process readings; without adverbials, activities and
progressive result in a continuous reading. Similarly, progressive futures
have to be marked adverbially precisely because they have neither an
obligatory process reading nor a repeated interpretation.
The use of adverbials, their role in communicating the intended
meaning of, and the interaction of the adverbials and the features of the
progressive support the expansion of the progressive from the prototypical
core to less prototypical readings. Römer presents an analysis of the
progressive that is dominated by two features: continuousness and
repeatedness. In the prototype association of progressive and activities the
feature of continuity is highlighted. In order to effect the spread from process
to repeatedness, adverbials are used. In both cases, the adverbs suppress the
expected reading of the progressive. In the repeated cases the adverbials
suppress the continuity feature, highlighting or introducing the repeated
feature. In the progressive future, the reading is neutral with respect to the
features of [continuousness] and [repeatedness].
The data in this longitudinal study suggest a modest refinement to
Andersen and Shirai’s proposal that the internal structure of the progressive
separates iterative from habitual. It seems that iterative and habitual are not
clearly distinguished in the production data. These can be grouped together as
repeated, as was done by Römer (2005) for native speaker corpus data and
Huang (1999) for L2 learner data. This suggests the following internal
structure for progressive:
Process (activity > accomplishment) > repeated (iterative / habitual) >
futurate > stative progressive.
A longitudinal investigation of the progressive prototype in L2 English 149
References
Hyun-Jin KIM
Indiana University
1. Introduction
The study examines the emergence and the spread of the Korean past /
perfective marker -ess- and the progressive marker -ko iss- in the L2 Korean
of L1 English speakers. Inspired by L2 Japanese studies on -te i- by Shirai
and his colleagues, the study investigates how L2 Korean learners associate
the progressive marker -ko iss-, which is comparable to the English
progressive be –ing, with a resultative meaning as well as a progressive
meaning. This study adopts the aspect hypothesis and the prototype account
to explain the development of the form-meaning association of the past /
perfective 1 and the progressive.
SLA research which addresses the relationship between the emergence
of tense and aspect morphology and the semantics of verbal predicates has
centred on the aspect hypothesis. The aspect hypothesis claims that the
distribution of interlanguage verbal morphology is determined by the lexical
aspectual class of verbs and predicates (Andersen & Shirai 1996; Bardovi-
Harlig 1998; see Li & Shirai 2000, and Bardovi-Harlig 2000 for recent
reviews). The aspect hypothesis consists of four predictions, which can be
summarized as “in the earliest stage, learners will use perfective / past with
achievements and accomplishments, imperfective with states, progressive
with activities, and no overextension of progressive marking to statives”
(Anderson & Shirai 1996: 559; Bardovi-Harlig 2000: 227).
Many empirical studies have addressed the aspect hypothesis and the
support for the first part of the hypothesis about the spread of the perfective /
past from telics to statives is well-established (summarized in Bardovi-Harlig
2000; see also Salaberry & Shirai 2002). With regard to the third claim of the
aspect hypothesis (in languages that have progressive aspect, progressive
marking begins with activities and later extends to accomplishments and
achievements), Bardovi-Harlig (2000) points out that “in cross-sectional
studies of English, progressive associates quite robustly with activities”
(237).
As an attempt to explain this early association of lexical aspects and
tense and aspect morphology, Shirai and Andersen (1995) proposed the
1
Past / perfectivity in Korean are marked by the morphology -ess-. Henceforth, I
will use ‘past’ instead of ‘past / perfective,’ for simplification.
prototype account. They suggested that the association between lexical aspect
and verb morphology in L1 and L2 acquisition can be attributed to the
acquisitional sequence from prototypical to peripheral members of the
linguistic categories of past tense and progressive aspect. In a prototype
account, learners infer a prototypical past as ‘completed action,’ and a
prototypical progressive as ‘action-in-progress’ 2. This action-in-progress
meaning is obtained “when the progressive marker is attached to activity
verbs and accomplishment verbs” (Shirai 2002: 457).
Research on the acquisition of tense and aspect morphology has been
done predominantly on Indo-European languages (Li & Shirai 2000;
Bardovi-Harlig 2000). However, Japanese is one of the few non-Indo-
European languages investigated as a second language within the framework
of the aspect hypothesis (Shirai 1995, 1999; Shirai & Kurono 1998; Sugaya
& Shirai 2007). Shirai and his colleagues’ work is important in that they have
drawn attention to Japanese by testing the universality of the aspect
hypothesis and also by testing crosslinguistic variation of the progressive /
durative marker -te i- within the prototype hypothesis. Japanese -te i- allows
both progressive and resultative interpretations depending on the lexical
semantics of the verb. Korean aspect marker -ko iss- behaves similarly to -te
i-, but interacts differently with the lexical semantics of the verb to which
they attach. The current L2 Korean research is drawing attention to the
acquisition of crosslinguistic variation of the progressive marker -ko iss- (e.g.
E.H. Lee & H. Kim 2007). These studies of Japanese and Korean will be
reviewed briefly in a later section.
This study is in line with the current research on L2 Japanese and L2
Korean tense and aspect acquisition. Before reporting the present case study
on the acquisition of past and progressive by learners of L2 Korean, the
following sections provide a brief introduction to Korean past and
progressive, and a review of the L2 research on the acquisition of tense and
aspect in Japanese and Korean.
2
Gass and Ard (1984) earlier suggested that on-going action is a prototype of the
progressive.
A case study of tense-aspect marking by L2 learners of Korean 155
relevant to the analysis carried out in the current study, the past tense -ess-
and the progressive aspect marker -ko iss- in Korean.
Korean verbal suffix -ess- [-ət]3 is a tense marker to denote past states or
events, which corresponds to the English past tense morphology -ed. The
core features of the past are [+past] and [+telic] (Shirai & Andersen, 1995).
Korean -ess- appears with all types of verbs, without any restriction in simple
sentences (Sohn 1995), with event verbs (1a), and with adjectival predicates,
as in (1b).
(1) a. Mary-ka ppang-ul muk-ess-ta.
Mary-NOM bread-ACC eat-PAST-DECL
‘Mary ate bread.’
b. Mary-ka kuttay kippu-ess-ta.
Mary-NOM then be glad-PAST-DECL
‘Mary was glad then.’
3
Korean past –ess- appears in allomorphs: -ess-[-ət] and -ass-[-at]
4
The conjunctive suffix –ko- ‘and’ carries a simultaneity feature, -iss- is the stem
of existential verb, issta.
156 Hyun-Jin Kim
Sentence (2) is ambiguous. With -ko iss- attached to the verb stem –ip-
‘wear’, this sentence can have either progressive or resultative meaning. To
express the meaning, “Mary is putting on the red dress” in Korean, both
“Mary-ka ppalkan os-ul ip-nun (NONPAST)-ta” and “Mary-ka ppalkan os-ul
ip-ko iss-ta” are possible. In contrast, in order to express the meaning “Mary
is dressed in red” in Korean, the second sentence with -ko iss- is acceptable,
5
As the comprehensive discussion of aspectual properties of these verbs in
Korean is outside scope of this paper, I will not pursue it in greater detail (but
see, Lee 1991; Kim 1993; Ahn, 1995).
6
Bybee et al. (1994) define resultative as “an aspect that signals that a state exists
as a result of past action” (54).
A case study of tense-aspect marking by L2 learners of Korean 157
but not the first sentence. There is a small set of verbs that behave this way.
They include verbs that are near equivalents of wear (3a) (which I will call
wear verbs), carry / hold (3b) (which I will call carry verbs) and body
posture verbs (3c).
(3) a. ‘wear’ verbs: ipta ‘wear clothes,’ sinta ‘wear shoes or socks,’
ssuta ‘wear glasses, hats’, kkita ‘wear gloves, ring,’
b. ‘carry’ verb: tulta ‘carry / hold,’ anta ‘hold in arms’
c. body posture verb: kamta ‘close eyes,’ ttuta ‘open eyes’
The fact that a single morpheme -ko iss- can express multiple meanings is of
particular interest to researchers working on Korean aspect, especially with
regard to the L2 acquisition of the form-meaning association, just as Japanese
-te i- has received much attention in the acquisitional literature. The question
is how we can account for this fact in Korean. Normally, the progressive
aspectual marker and resulting state aspectual marker are grammaticized in
different morphemes in many languages (Bybee, Perkins, & Pagliuca 1994).
For example, Korean has two aspectual morphemes; -ko iss- for the
progressive and -a iss- for the resultative marker. However, their boundary is
not clear as illustrated in the example (2) with -ko iss-. K. Lee (1978 cited in
Shirai 1998: 666) suggests that “the resulting states expressed by the
progressive -ko iss- are those initiated and maintained consciously by the
agent.” Shirai (1998) added that if a resulting state is viewed as involving
dynamicity / agency, the progressive -ko iss- can enter the territory of
resulting state (as in the case of verbs in (3)). The differences between
Korean and Japanese aspectual markers reflect the results of the different
patterns and degrees of grammaticization of their imperfective marker (Shirai
1998).
The next section provides a brief review of L2 research addressing this
interesting issue of Japanese and Korean aspectual markers, as well as the
aspect hypothesis.
Shirai and Kurono (1998) tested the aspect hypothesis with L1 Chinese
learners of L2 Japanese, using an oral production task and an acceptability
judgment task. They found that learners show strong association of past
marking -ta with achievements, and of progressive marker -te i- with
activities. They also suggest that the progressive use of -te i- with activities
appears earlier than its resultative state use with achievements, indicating that
progressive meaning is the prototype of -te i- in L2 Japanese. Sugaya and
Shirai (2007) examined the influence of learners’ L1 on the acquisition of
Japanese -te i-using an acceptability judgment test and an oral description
158 Hyun-Jin Kim
task. The results from the acceptability judgment test support the aspect
hypothesis, in that -te i- was strongly associated with activities for lower
proficiency learners, regardless of L1. However, the results from the oral task
did not support the prediction, in that lower proficiency learners from
German and Slavic L1 background did not show any preference. They claim
that L1 plays a role in the formation of the acquisition pattern predicted by
the aspect hypothesis, but L1 influence is not the only possible answer
available to explain the acquisition of the Japanese aspect.
With the exception of research in Japanese mentioned previously (Shirai
2002; Shirai & Kurono 1998; Sugaya & Shirai 2007), very few studies have
been conducted in non-Indo-European languages within the framework of the
aspect hypothesis, and L2 Korean has received little significant research
attention within the framework of the aspect hypothesis. However, a notable
exception is E.H. Lee and H. Kim’s (2007) study of the acquisition of two
Korean aspect markers (the progressive -ko iss- and the resultative -a iss-) in
L2 Korean by learners with various L1s. They examined how aspectual
morphology expands beyond prototype associations with inherent lexical
aspects. They hypothesized that -a iss- will develop later than -ko iss- for
progressive, but earlier than -ko iss- for resultant state. In the cross-sectional
data from a sentence interpretation task and a guided picture description task,
they found the early emergence of -ko iss- with action in progress meaning,
and they also found that the order of development of -ko iss- action precedes
-a iss- in resultant states. The results supported their hypothesized acquisition
order of two Korean aspect markers, with individual variation.
The study examines the emergence and the spread of the Korean past tense
marker -ess- and the progressive marker -ko iss- in English-speaking L2
Korean learners’ longitudinal production data. Bardovi-Harlig (2000)
suggests that the universality of the aspect hypothesis be tested with a wide
range of the target languages. To that end, this case study addresses the
aspect hypothesis in L2 Korean, which has not previously been investigated
within this framework. In addition, the study investigates how L2 Korean
learners associate the progressive marker -ko iss- with resultative meaning as
well as progressive meaning (cf. E.H. Lee & H. Kim 2007; Shirai & Kurono
1998; Sugaya & Shirai 2007). Given that little attention has been paid to
cross-linguistic variation in the aspect hypothesis, this study should be a
meaningful attempt to understand how tense and aspect morphology expands
to nonprototypical associations over time with relation to lexical aspects.
Specifically, the study addresses the following two research questions:
A case study of tense-aspect marking by L2 learners of Korean 159
3.1. Method
3.1.1. Participants
The participants in this study are two native speakers of English (Kate and
Beth) who were enrolled in a Korean language program at an American
university 7. They were enrolled in the third semester of Korean (first
semester of second-year Korean) at the beginning of this study and
progressed to the fourth semester. They had had no precollege instruction in
Korean. They are both females in their early twenties. Students completed a
language background questionnaire prior to the beginning of data collection.
In their Korean program curriculum, the past tense is introduced in the first
semester of first-year Korean, and the progressive meaning of -ko iss- is
introduced in the second semester of first-year Korean. The result state
meaning of -ko iss- is introduced later when they learn how to describe
people with wear verbs. By the time of data collection, the past and
progressive forms had been already taught to these learners.
The present study was conducted over a period of six months from November
2006 to April 2007. The researcher met individually with each participant to
record samples of oral production. The study consists of six data collection
sessions at three to four-week intervals. The first two data collection sessions
were administered when the participants were in the third semester in Korean
program. Further data were collected when they were in the fourth semester.
Each session was approximately 45 minutes in length. The audio data were
recorded digitally on the computer using the Audacity program while the
participants were performing the tasks, and were then transcribed by the
researcher using Korean orthography.
3.1.3. Tasks
In each session, the participants were engaged in two tasks which elicit
spontaneous oral production: a conversational interview and a picture
description task. The conversational interview targeted past contexts with
7
The number of participants is due to the limited availability of L1 English
students in the second year of the Korean program in this university.
160 Hyun-Jin Kim
3.2. Analysis
The data coding procedures basically follow those of L2 English studies (e.g.
Bardovi-Harlig, 1995, 1998), but were modified to fit Korean linguistic
structures. The data were analyzed for each task.
Only the verbs in the past time context were coded and token analysis was
conducted. First, each predicate or verb phrase was identified and classified
into one of Vendler’s (1967) four lexical categories. If the same verb
expresses different semantic relationships to its object (Dowty 1979), it
belongs to a different aspectual class. The lexical aspect of each predicate
was determined by means of established operational tests (Dowty 1979;
Shirai & Anderson 1995), and relevant modification was applied to the
8
In order to elicit present and progressive, the researcher showed the pictures to
learners and explained in English at the beginning, “This is happening now and
you are watching this scene, but I cannot see the scene. I would like to know
what is going on now. Please explain to me what you see in the picture and what
is happening now in detail.” The researcher could not see the pictures.
A case study of tense-aspect marking by L2 learners of Korean 161
9
For example, cognitive verbs that are classified as statives in English were
classified as achievement in this study (e.g. know, as discussed in section 2.2).
10
Within-category analysis and across-category analysis differ in reporting rates
of use of morphology in each aspectual category, and in the sensitivity to
number of tokens produced (Bardovi-Harlig 2000). Across-category analysis is
so sensitive to the number of predicate tokens in each category that the
percentages of predicate use in each category can be easily inflated by this
analysis. In contrast, within-category analysis does not address the actual
number of tokens, but rather capture the rate at which a learner can inflect
activities with the morphology such as the simple past.
162 Hyun-Jin Kim
Data coding procedures of the picture description task were basically the
same as a conversational interview. First, each predicate was identified and
classified into one of four lexical categories. Second, all predicates were
coded for verbal morphology: past (-ess / ass-), progressive (-ko iss-), and
nonpast (-e / ayo). Then, the percentage of use for verbs from all lexical
aspect categories marked by the progressive was calculated and compared
between participants. The predicates were also coded for the corresponding
uses as progressive or resultative state. All resultative contexts with -ko iss-
involved wear shoes (dress, glasses, hats), carry, and hold, 11 which were
coded as wear / carry verbs.
3.3. Results
3.3.1. Conversational interview
The results from the six conversational interviews clearly show that the
learners’ use of the past tense marker -ess- was associated with achievements
and accomplishments in the initial phase. Statives tended to be used with the
nonpast in the interlanguage of both learners. The use of -ess- with statives
was low. Adjectival predicates, a subclass of statives, also show the same low
distribution with regard to -ess-. In the later sessions, the spread of past from
events to states was observed in both learners’ production.
Distribution of verbal morphology. The conversational interview yielded an
average of 31 tokens (138 tokens / 6) per session in Kate’s past-time
reference and 29 tokens (172 tokens / 6) in Beth’s. This section reports the
results from each participant separately.
Table 2 presents Kate’s distribution of verbal morphology within the
aspectual categories in past contexts. Examples of each category from Kate’s
data are found in examples (6a)-(6d). The session number is indicated in the
parenthesis.
(6) a. ACH: Sang bat-ass-eyo. “I received an award.”(S6)
b. ACC: Chapchae meok-ess-eyo. “I ate Chapchae (a Korean dish)” (S5)
c. ACT: Nun-i ow-ass-eyo. “It snowed.” (S3)
d. STA: Haengbokha-ess-eyo. “I was happy.” (S4)
11
Stative predicates “is seated, is open, is closed,” are also in the resultative
context, but they are obligatorily marked by -a iss- for resultative meaning.
Since -a iss- is not a focus of the current study, learners’ production of these
three stative-type verbs will not be discussed in detail.
A case study of tense-aspect marking by L2 learners of Korean 163
The results from oral interviews showed patterns that have been well-
supported in studies on the aspect hypothesis. Over the course of the six data
collection periods, Kate showed the greatest use of simple past marking on
achievements, followed by accomplishments. She showed the lowest use of
past with statives as predicted in the aspect hypothesis. As shown in Table 2,
the only competitor to the past form in every aspectual category was the
nonpast. No progressive occurred in this task.
Kate’s developmental pattern over time is illustrated in Figure 1. The
use of the past marker with stative verbs increased noticeably from 50% (5
out of 10 stative tokens in Session 1) to 93% (out of 6.5 tokens in Session 6)
throughout the six sessions. The use of the past with accomplishment verbs
also increased from 50% (out of two tokens in Session 1) to 100% in Session
4, 5 and 6. Achievements show the highest rate of past tense marking and this
pattern persisted over time (88% - 100%). Unstable use of the past with
activity verbs was observed between Session 1 and Session 4. Later on, it
steadily increased and reached 100% out of 7 verb types in the last session.
164 Hyun-Jin Kim
In sum, use of the past with achievements and accomplishments and its
spread to activities is observed in Kate’s production over time. The
successful spread of the past to states is also observed. Even though the
number of coded verb tokens in each category was sometimes small (two or
three tokens), the study could trace the overall pattern of her development in
supplying the past marker -ess-. By the end of the study, Kate could move
appropriately from past to present, and from present to past, according to the
conversational contexts.
Beth’s pattern of associating the past with four lexical categories is
similar to Kate’s, but is evidenced to a different degree. From the beginning
of the study, Beth was less advanced than Kate with regard to past tense
marking. Table 3 displays Beth’s distribution of verbal morphology.
Examples of each category are found in examples (7a)-(7d).
(7) a. Achievement: Pul na-ass-eyo. “A fire broke out.” (S4)
b. Accomplishment: Yeonghwa po-ass-eyo. “I watched a movie.” (S4).
c. Activity: Chumchuw-ess-eyo. “I danced.”(S2)
d. Stative: Wonhae-ess-eyo. “I wanted (TV).” (S6).
A case study of tense-aspect marking by L2 learners of Korean 165
Figure 2. Distribution and spread of the simple past in Beth’s conversational interview
In sum, the use of past with events and its spread to activities is
observed in Beth’s oral sample over time. In contrast to Kate, who achieved
93% use of past by Session 6 with statives, the spread of past to states is not
clearly demonstrated in Beth’s data.
The acquisition data from both learners revealed a strong association
between past tense and achievements and accomplishments with individual
differences in the strength of the association. The data also show strong
association of statives and adjectival predicates with the nonpast. The spread
of past from events to states in the later sessions was clearly observed in
Kate’s production and to a lesser extent in Beth’s.
In this section, the results from a picture description task by each participant
will be reported separately. At the time of the task, the progressive had
emerged in both participants’ interlanguage, but with different patterns and to
different degrees. The progressive began with activities and accomplishments
and spread to achievements, especially in Kate’s data. Neither learner used
progressive forms with stative verbs, which means that the progressive is not
being overgeneralized, consistent with the aspect hypothesis.
Distribution of verbal morphology within aspectual categories. The
distribution of the tense and aspect morphology within aspectual categories
by Kate is provided in Table 4.
A case study of tense-aspect marking by L2 learners of Korean 167
At the time of its emergence, -ko iss- was associated with activity verbs
and accomplishment verbs. In the following session, it had spread to
achievements. In the final session, she showed 77% use of progressive with
activities and 67% use with accomplishments.
The distribution of the tense and aspect morphology within aspectual
categories by Beth is provided in Table 5.
A case study of tense-aspect marking by L2 learners of Korean 169
I will now examine -ko iss- in progressive and resultative contexts more
closely. To describe the given pictures, ‘wear’ and ‘carry’ are expected to
appear with the progressive morphology. They are in the preferable context
of -ko iss- to have the resultative meaning (e.g. “An old lady is wearing a
white dress.”)
(9) *Halmoni hayan one-piece ip-eyo.
Old lady white dress wear- Pol.
‘An old lady wears a white dress.’
In order to express the meaning “An old lady is wearing a white dress”
in Korean, -ko iss- is needed as in “Halmoni-ka hayan one-piece-lul ip-ko
iss-eyo.” Table 6 illustrates the path of form-meaning association of the
progressive with wear / carry verb tokens over time when learners attempted
to express resultative state meaning in the pictures.
Table 6. Verb-tokens of wear / carry by learners
S1 S2 S3 S4 S5 S6 Total
Kate Progressive 0 0 0 0 9 7 16
Nonpast 2 6 7 5 6 3 29
Past 0 0 0 0 0 0 0
Beth Progressive 0 0 0 0 0 0 0
Nonpast 2 4 1 1 9 8 25
Past 0 0 6 2 9 6 23
From Session 1 through Session 4, Kate never showed the use of the
progressive marker with ‘wear / carry’ verbs. It is assumed that she did not
associate -ko iss- with the resultative meaning until Session 5. However, note
that nine verb tokens with -ko iss- for the resultative state meaning emerged
in Session 5. In Session 5, she produced four verb types with -ko iss- for the
progressive meaning (yekiha-ass-eyo ‘talked’→yekiha-ko iss-eyo ‘is talking’,
ja-ko iss-eyo ‘is sleeping’, shoppingha-ko iss-eyo ‘is shopping’, and manna-
ko iss-eyo ‘is meeting’), and five verb types with –ko iss- for the resultative
state meaning (ssu-ko iss-eyo, sin-ko iss-eyo, tul-ko iss-eyo, kutu ssu-ko iss-
eyo, yangbok ip-ko iss-eyo). Interestingly, they were self-repaired from
nonpast to progressive (the target form) in the resultative contexts, as in
example (10).
(10) a. Moja ssu-eyo…uh Moja ssu-ko iss-eyo.
hat wear-nonpast, hat wear-ko iss-Pol.
‘(She) wears a hat, uh, is wearing a hat.’
A case study of tense-aspect marking by L2 learners of Korean 171
In Session 6, Kate used six verb types with -ko iss- for the progressive
meaning, and five verb types with -ko iss- for the resultative state meaning.
In this session, she directly produced the progressive form for six cases as in
ip-ko iss-ta, and ssu-ko iss-ta, and she repaired only once from nonpast to
progressive (kky-eyo → kky-ko-is-eyo).
In the last two sessions, Kate attempted to use appropriate verbs which
are compatible with the direct objects in order to express the meaning of
‘wear.’ In Session 5 and 6, she used, ipta ‘wear (clothes),’ sinta ‘wear (shoes
or socks),’ and ssuta ‘wear (hats).’ Example (11) shows a lexical search for
the target verb (sinta) and repair to the target form (sinkoissta) at the same
time. This utterance is interesting to observe, considering that Korean has
different verbs for ‘wearing shoes,’ ‘wearing clothes,’ and ‘wearing glasses.’
(11) Miss Fanny kutu ip-eyo, sin-eyo, sin-ko iss-eyo.
Miss Fanny shoes wear-nonpast, wear-nonpast, wear-ko iss-Pol
‘Miss Fanny wears (for clothes), wears (for shoes), is wearing shoes.’
Overall, the results from Kate’s picture description task showed her
development of the association of -ko iss- with progressive and resultative
meaning over time. In terms of the emergence of -ko iss-, progressive
meaning emerged earlier than resultative meaning.
Beth’s developmental pattern is significantly different from Kate’s
regarding the progressive marking. As illustrated in Table 6, Beth never
showed the use of the progressive marker with ‘wear / carry’ verbs across all
six sessions. She had not associated -ko iss- with the resultative meaning. The
results from Beth’s picture description task may suggest that the progressive
meaning of -ko iss- emerged earlier (one token of ACT at Session 4, see
Table 5) than its resultative meaning which had not appeared yet. However,
given that the progressive meaning of -ko iss- appeared only once, it is
premature to make any strong claim about Beth’s development of the form-
meaning association of -ko iss- at this point.
Alternatives to progressive and resultative. To describe their interlanguage in
relation to the aspect marking in detail, I examined the alternatives to the
progressive used by Kate and Beth. The use of -ko iss- and other alternative
forms to denote progressive and resultative state meaning by learners is
provided in Table 7.
172 Hyun-Jin Kim
4. Discussion
4.1. Does the aspect hypothesis correctly predict development in L2
Korean past and progressive marking?
The study also suggests that -ko iss- was used for action-in-progress meaning
first with activities before it was used for resultant state meaning with Korean
wear / carry verbs. Even though it appeared in only one learner’s production,
the results from the picture description task showed the developmental
pattern in L2 Korean learners’ association of -ko iss- with wear / carry; the
progressive meaning of -ko iss- emerged earlier than its resultative meaning.
This result is mainly consistent with previous studies which address the
progressive and resultative meaning of this aspect marker (E.H. Lee & H.
Kim, 2007 for L2 Korean; Shirai and colleagues’ work for L2 Japanese).
As for the progressive marking, the results are also consistent with the
prediction of prototype account. Kate’s results can be interpreted as the
spread of a prototypical association: from action in progress to resultative
state. It is the extension from prototypical progressive with the feature of
[−telic] and [+dynamic] (Shirai & Andersen, 1995; Shirai, 2002) to
resultative with [+durative] and [−dynamic] which is nonprototypical. In
Korean, resultative -ko iss- is less prototypical than progressive -ko iss-. It
174 Hyun-Jin Kim
seems that the feature mismatch between the progressive form and the
resulting state meaning ([+dynamic] versus [−dynamic]) creates a difficulty
in its form-meaning association.
Given that Korean -ko iss- with resultative meaning is not as productive as
Japanese -te i-, a learnability question might need to be raised. How can
learners acquire knowledge about the interaction between the semantics of
the morphological forms and restricted sets of words ? From the input they
receive, it is highly likely that learners perceive the progressive -ko iss- with
activities or accomplishments more frequently than the resultative -ko iss-
with wear / carry verbs. As for the classroom instruction which learners in
this study are mainly exposed to, -ko iss- with the progressive meaning is
taught first and given serious attention within the grammar syllabus, whereas
-ko iss- with the resulative meaning is introduced as lexical expression to
describe people using wear verbs. Since the distinction between the result
state meaning and the progressive meaning of -ko iss- is not explained
explicitly in the instruction, the multifunctionality of -ko iss- might pose a
difficulty for the learners in acquiring both meanings equally, by moving
from the initial one-to-one association of -ko iss- to activities. Yet, it is also
possible for learners to receive naturally occurring input from their instructor,
from their peers, or from the media. This issue could be investigated by
further examining corpus data in the future research. Multiple factors seem to
be involved when learners acquire the subtle interplay between the aspect
marker and the semantics of verbs. Further research employing different
methodologies is needed to investigate this complexity.
not use the progressive as they would in their L1. In Session 3, Kate switched
from Korean to English when she was describing a female character in the
picture, showing that she would add progressive in her L1 production to
describe the picture, but she didn’t do so in her L2.
(12) Halmoni chinku-ka sesuha-eyo, halmony towu-ayo.
She is helping.
Old lady friend-NOM wash face-Pol, old lady help-Pol
‘Her friend washes her face, helps her.’ (in Korean) “She is helping” (in
English)
5. Conclusion
This study suggests that the results support the applicability of the aspect
hypothesis to L2 Korean. Even though it followed only two learners’
development, this case study provides some evidence that the aspect
hypothesis correctly predicts the development of L2 Korean tense and aspect.
In addition, data collection methods were less guided than other judgment
tasks, emphasizing the importance of natural conversation data in testing the
aspect hypothesis. The longitudinal design allowed to keep track of the
learners’ emergence and spread of tense and aspect markers over time in a
qualitative way. Yet, it is possible that a cross-sectional study with larger
learner samples may yield a different perspective on the acquisition of -ko
iss- with a progressive meaning and a resultative meaning. For future
investigation in this regard, a finely tuned elicitation task design will be
necessary.
176 Hyun-Jin Kim
References
Ahn, Y. (1995). The aspectual and temporal system of Korean: From the
perspective of the two-component theory of aspect. Unpublished
doctoral dissertation, University of Texas at Austin.
Anderson, R.; Shirai, Y. (1994). Discourse Motivation for some cognitive
acquisition principles. Studies in Second Language Acquisition 16: 133-
156.
Bardovi-Harlig, K. (1998). Narrative structure and lexical aspect: Conspiring
factors in second language acquisition of tense-aspect morphology.
Studies in Second Language Acquisition 20: 471-508.
Bardovi-Harlig, K. (2000). Tense and Aspect in Second Language.
Acquisition: Form, Meaning and Use. Malden, MA: Blackwell
Bardovi-Harlig, K. (2002). Analyzing aspect. In M. R. Salaberry; Y. Shirai,
(eds), The L2 acquisition of tense-aspect morphology. Amsterdam:
Benjamins, 129–154.
Bardovi-Harlig, K.; Reynolds, D.W. (1995). The role of lexical aspect in the
acquisition of tense and aspect. TESOL Quarterly 29: 107-131.
Bybee, J. L., Perkins, R.; Pagliuca, W. (1994). The evolution of grammar:
Tense, aspect and modality in the languages of the world. Chicago: The
University of Chicago Press.
Dowty, D. (1979). Word meaning and Montague grammar. Dordrecht:
Reidel.
Gass, S. M.; Ard, J. (1984). Second language acquisition and the ontology of
language universals. In W.E. Rutherford, (ed.), Language universals
and second language acquisition. Amsterdam: Benjamins, 33-68.
Kim, Y. (1993). The resultative progressive in Korean. Papers of the 29th
Regional Meeting. Chicago Linguistics Society: 251-265.
Lee, E. H. (2006). Stative progressives in Korean and English. Journal of
Pragmatics 38: 695-717.
Lee, E. H.; Kim, H. (2007). On cross-linguistic variations in imperfective
Aspect: The Case of L2 Korean. Language Learning 57: 651-685.
Lee, H. S. (1991). Tense, aspect and modality: A discourse-pragmatic
analysis of verbal suffixes in Korean from a typological perspective.
Unpublished doctoral dissertation, UCLA.
Lee, K. (1993). A Korean Grammar on Semantic-pragmatic Principles.
Seoul, Korea: Hankwuk Mwunhwasa.
Li, P.; Shirai, Y. (2000). The Acquisition of Lexical and Grammatical Aspect.
Mouton de Gruyter: Berlin.
Salaberry, R.; Shirai, Y. (2002). The L2 Acquisition of Tense-Aspect
Morphology. Amsterdam: Benjamins.
A case study of tense-aspect marking by L2 learners of Korean 177
Evgeniya SERGEEVA
Jean-Pierre CHEVROT
Laboratoire LIDILEM – Université Stendhal Grenoble 3
1. Introduction
inflected verb forms and their stems. Stemberger (2004) and Tabak (2005)
have obtained evidence of competition between the past and present tense
forms of English verbs.
The usage-based theories (Langacker 1987; Bybee 2001) seem to
provide a better account of the acquisition of stem allomorphy. These
theories postulate that such acquisition is based on exemplars which are
memorized and linked to one another. This memorization leads to the
emergence of schemata with variable levels of abstraction (Ellis 2002).
Moreover, the usage-based theories are particularly appropriate for
addressing our research issues given that, in contrast to the dual-route
models, they predict relations between stem allomorphs. According to Bybee
(1985, 1991), some inflectional forms can be described as "the stem of one
form + a marker." Bybee points out that this kind of relation is also possible
for stem alternations. Bybee claims that children and L2 learners start
acquiring conjugations by using one verb form to replace all the others. This
would usually be the base form of the paradigm. All the other forms could
then be derived from the base form which is frequent and semantically non-
marked. Learners then subsequently acquire the functions of the other forms
and use the base form for building needed forms by adding suitable markers.
In most cases, the base form is the singular present indicative.
The aim of the experiment was to observe the production of conjugated forms
of French verbs, paying particular attention to the selection of verbal stems.
Our objective was to define the role of different factors, and in particular the
role of frequency and paradigm structure, in the observed productions and
errors.
To this end, we developed a verbal elicitation task which included verbs with
different models of stem alternation. This classification of verbs in terms of
models of stem alternations and frequencies allowed us to choose three verbs
with high cumulative frequencies (faire /to make/, prendre /to take/, devoir
/to have to/) and three with low cumulative frequencies (boire /to drink/,
craindre /to fear/, traduire /to translate/) (Table 2).
The temporal and modal forms to be elicited were selected on the basis
of the grammatical knowledge of the Russian learners who participated in the
task. An analysis of textbooks used during their French L2 classes allowed us
to identify the tenses and moods that they had learned and those that they had
not. The elicited forms were: the présent 3sg., 1pl., 3pl., the subjonctif
The acquisition of French verbal tenses by Russian adult learners 183
présent 3sg., the futur simple 3sg., the participe passé and the infinitif. These
tense-person forms were chosen due to the fact that they represent the range
of variants of verbal stems.
Table 2. Elicited verbs and their stem alternations
Cumulative frequency
Group of forms
More frequent verbs Less frequent verbs
faire prendre devoir boire craindre traduire
imparfait, présent
A/B A A A A A
1&2pl.
présent 3 pl. C B B B A A
présent sing. D C C C B B
subjonctif présent
E A A A A A
1&2pl.
subjonctif présent
E B B B A A
sing., 3pl.
impératif sing. D C C C B B
impératif pl. A A A A A A
participe présent A A A A A A
futur-conditionnel F D A C C B
passé simple G E D D D C
infinitif D D E C C B
participe passé D E D D B B
4.1.2. Participants
4.1.3. Procedure
All the subjects completed the task individually. Each target form was
elicited twice in two different conditions. In the implicit condition the
participants had to complete an utterance prompting the target form by an
184 Evgueniya Sergeeva & Jean-Pierre Chevrot
adaptation of the Berko-test (Berko 1958). For example, when the target form
was the futur simple 3sg., the experimenter says “aujourd’hui Marc boit du
jus et demain encore il…” [today Marc drinks some juice and tomorrow
again he …]. The participant was expected to produce “il boira” [he will
drink].
In the explicit condition, the infinitive form of the verb was given to the
participants as well as the grammatical term identifying the target form. For
example, “boire, 3ème personne du futur” [to drink, futur simple 3sg.]. It
should be noticed that in French the infinitive form is the only way to refer to
a verb out of the context of an utterance 1.
The verbs and tense-person forms were presented in random order. The
instruction was presented orally and the responses were recorded.
4.1.4. Predictions
1
Comparing these two conditions we aimed to bring out the part of the
metalinguistic reasoning which was expected to be more prominent in the
explicit condition. Given that no significant difference distinguishes the
performance rates in these two conditions, the distinction between them will not
be taken into consideration in the following results.
The acquisition of French verbal tenses by Russian adult learners 185
To our knowledge, few studies have attempted to establish the link between
frequencies in the real input of L2 learners and the output produced by these
same learners. However, as Chernigovskaya and Gor (2003) noted, instructed
L2 learners have limited access to verbal frequencies when compared to
children learning their L1. These authors suggested performing a frequency
count based on the classroom textbooks used by the American students of
Russian who participated in their experiments. In the same way as for the
studies of the acquisition of French L2, Thomas (2005) produced a frequency
count based on the teacher’s utterances in a corpus of guided interviews.
Gor and Chernogivskaya (2003) analyzed the productions of beginner
learners of Russian. In contrast, the learners in our experimental groups had
received at least 2 years of French tuition. Moreover, the input survey
showed that they had sources of input other than their French lessons.
Therefore, a frequency count derived from a corpus based on the French
classroom textbooks reflects only part of the input. We will, however, use the
verb frequencies in this corpus and compare them to those listed in the
Lexique 3 database (New 2001).
In our study, all the students were learning French with the Nouveau
sans frontières package (NSF). The learners in the LP group had completed
two levels of the package, whereas the learners in the HP group had
completed four levels. The textbooks and workbooks were scanned in full,
after which we ran optical text recognition software in order to obtain a
digital corpus. We analyzed each page in full with the exception of page
numbers and titles. The frequency count was applied to the forms written in
the textbooks and workbooks. The forms that had to be produced in the
exercises were deliberately omitted. A Pearson correlation test was used to
calculate the correlations between the frequencies listed in Lexique 3 and the
frequencies obtained from our corpus. The indices show a high level of
correlation (Table 3).
Table 3. Correlation of frequencies obtained from the analysis
of the NSF corpus and the frequencies listed in Lexicon 3 (New, 2001)
Correlation coefficient
Unit of frequency count
NSF – Lexique 3
Target inflected form 0.892 (p<0.001)
Target form homophones 0.990 (p<0.001)
Target stem 0.989 (p<0.001)
This result means that the overall frequency of verbal forms is very well
represented by their textbook occurrences.
186 Evgueniya Sergeeva & Jean-Pierre Chevrot
4.3. Results
The analysis of the results follows the same order as the predictions we made
concerning the frequency and base forms effect in the production of the
verbal forms of French L2.
Table 4 shows the mean rates of the different stem/inflection
combinations as a function of their accuracy.
Table 4. Overall rates of different stem/inflection combinations
Stem-inflection combination Rate Example
Correct inflected form 57.5% /il bwara/
Incorrect stem + correct inflection 28.8% /il byvra/
Correct stem + incorrect inflection 3.8% /il bwarE/
All incorrect 9.9% /il byvrE/
Total 100%
It appears that most errors refer to the choice of stem, while the
inflection is selected correctly. This result is consistent with previous
findings that, for learners of French L2, the tense marking expressed by a
stem choice is more difficult to acquire than the marking expressed by a non-
zero inflection (Ågren 2005; Bartning 2004; Granget 2005; Hedbor 2005;
Schlyter 2005).
Some of these authors formed the hypothesis that the acquisition of
stem alternation in French L2 could be biased by the frequency effect
(Schlyter & Bartning 2005, Schlyter 2005). Only few studies relate real input
frequencies to the productions of the French L2 learners (e.g. Thomas 2005).
To our knowledge no study, however, has tried to manipulate the frequency
in a task prompting the production of verbal forms by learners of French.
2
We chose not to exclude the non-responses from the total number of forms
elicited in the task, since they represented only 0.6% of the total number of the
forms elicited in the task. The mean accuracy rates were calculated using the
following formula:
Number of correct forms / stems
Total number of forms in the task – Number of non-responses
The acquisition of French verbal tenses by Russian adult learners 187
Table 5. Accurate production of inflected forms and stems of high and low-frequency
verbs (mean rate and standard deviation)
Rates of correctly produced forms Rates of correct stem choice
High- Low- High- Low-
t-test t-test
frequency frequency frequency frequency
significance significance
verbs verbs verbs verbs
57.88 45.08 t=2.2609 88.22 64.22 t=4.0267
(27.39) (25.39) p=0.0275 (21.19) (15.28) p<0.001
The table shows that the mean rates of correct production were higher in
high-frequency verbs. These results are significant both for the production of
complete verbal forms and for stem selection. The influence of frequency on
the production of verbal morphology has been largely documented in second
language research (e.g. Ellis & Schmidt 1997; Brovetto & Ullman 2001;
Birdsong & Flege 2001; Gor & Chernogovskaya 2003; Kirkici 2005). Our
results show that this effect is also present in Russian speaking learners of
French.
For the learners in the LP group, the correlations between the correct
form rates and both the surface frequencies and stem frequencies were
significant. We also found significant correlations between the correct stem
choice and the frequencies of the forms that were built with these stems and
the frequencies of the stems themselves.
In the HP group, we observed strong and positive correlations between
the accuracy rates of the forms and their surface frequencies, as well as
between the accuracy of forms and the frequencies of their stems.
Nevertheless, there was no significant correlation between the rates of correct
stem selection and the frequencies of the form or the frequency of the stem
itself.
In order to see whether there was a change in the frequency effect, the
correlations that were significant in both groups of learners were compared. It
appears that the correlations in the HP group were stronger for the link
between the effect of the frequency of the target forms (p = 0.033) and stems
(p = 0.047) and the accuracy rates of the inflected forms. Overall, frequent
forms and stems are likely to be produced correctly and this frequency effect
increases with the learners’ proficiency.
We observed that the more advanced learners (HP group) produced the
frequent verbs more accurately than verbs with low cumulative frequencies.
Moreover, the correlations show that link between the frequency and the
accuracy is stronger in the HP group. These results are in line with the idea of
Ellis (2002) who underlines that learners need have processed sufficient
exemplars for their experience to be representative of the content and the
frequencies of the target language.
The acquisition of French verbal tenses by Russian adult learners 189
3
All the frequencies are given per million words.
190 Evgueniya Sergeeva & Jean-Pierre Chevrot
pattern). For this error type, the difference was significant for the comparison
of both forms (t=8.376, p<0.001) and stem frequencies (t=7.317, p<0.001).
When a simple form replaces a complex form (SF for CF pattern), the
substitute is more frequent in terms of both form frequency (t=4.668,
p<0.001) and stem frequency (t=3.820, p=0.031). In the case of the CF for
CF pattern, we found no significant difference between the frequencies of the
target and substitute forms with complex structures (t=1.663, p=0.098).
However, we found that the substitute stem of a complex form is more
frequent than the target stem of other complex forms (t=4.014, p<0.001). In
cases where a complex form replaced a simple form (CF for SF pattern), the
difference between the form frequencies was not significant (t=0.663,
p=0.508) and neither was that between the stem frequencies (t=1.432,
p=0.154).
Generally, more frequent forms and stems are likely to replace less
frequent ones. This frequency effect is more pronounced when the substitute
has a simple structure.
Table 10. Correct forms and stems as a function of target forms and proficiency levels
(mean rates and standard deviations)
Target Correct form rates Correct stem rates
form LP HP LP HP
3PR 74.44 (16.51) 92.20 (8.59) 75.45 (16.51) 92.23 (8.60)
INF 65.55 (16.87) 95.54 (16.92) 65.57 (27.08) 95.56 (7.62)
PP 64.42 (23.71) 84.98 (14.84) 64.46 (22.82) 85.01 (15.17)
4PR 44.44 (21.10) 70.54 (25.29) 45.56 (17.78) 70.01 (16.91)
3FUT 23.89 (17.51) 62.77 (25.17) 43.89 (21.25) 70.57 (17.50)
6PR 22.22 (22.82) 65.00 (15.17) 27.22 (22.60) 66.11 (14.94)
3SUBJ 21.12 (27.08) 52.22 (7.62) 24.99 (19.92) 52.78 (24.13)
3PR – présent 3sg., INF - infinitif, PP – participe passé, 4PR – présent 1pl.,
3FUT – futur simple 3sg., 6PR – présent 3pl., 3SUBJ – subjonctif 3sg.
In this table, we observe that three forms - 3PR, INF and PP (Forms1) -
differ from the others (Forms2) in exhibiting higher rates of accurate
production. We checked whether the accuracy rates were statistically
different in the Forms1 group.
In the productions of the LP learners, there was no significant difference
between the accuracy rates for 3PR and INF (t = 1.086, p = 0.287), for PP
and 3PR (t = 1.378, p = 0.178), or for PP and INF (t = 0.123, p = 0.903). The
accuracy rates on the target forms in the Forms1 group (3PR, INF, PP) were
systematically higher than the rates in the Forms2 group (4PR, 6PR, 3SUBJ,
3FUT) (p<0.01).
An almost identical pattern was found in the productions of the HP
learners. No significant difference was found between the rates of correct
production of inflected forms of 3PR and INF (t = -1.123, p = 0.271), or
between PP and 3PR (t = 1.603, p = 0.120). The difference was significant
between PP and INF (t = -2.408, p = 0.022). As in the productions of the LP
learners, the accuracy rates for forms in the Forms1 group were higher than
in the Forms2 group (p<0.02).
Overall, the three forms considered in this study were distinguished by
high rates of production accuracy at both levels of proficiency. This implies
that these forms are less prone to error than other elicited forms, even in the
productions of LP learners.
In order to check whether the high accuracy rates of these forms were
frequency-based, we performed bivariate correlation tests between the
accuracy rates and frequencies of the forms and the stems for each target
form and for each group of subjects. The summary of the correlation
coefficients is presented in Table 11.
We found strong and positive significant correlations between frequency
and accuracy for three forms in the Forms2 group (4PR, 6PR and 3SUBJ),
thus indicating that the greater the surface frequency of these forms, the
better they are produced. Interestingly, the frequency - performance
192 Evgueniya Sergeeva & Jean-Pierre Chevrot
frequency
frequencies (* - the correlation is significant at the 0.05 level)
It is not difficult to see that Russian learners use the 3PR to build the
verbal paradigm. In contrast, the use of the INF is more ambiguous because
of a low occurrence of substitution errors. To observe how the INF was used
to build the verbal paradigm, we performed a more detailed analysis of the
stem choice errors built with existent stems. In all the cases when a substitute
infinitif stem was used, the three most frequent targets were the 3FUT (19/70
replacements by an infinitif stem: il /dəvwara/ replaces il /dəvra/), the 6PR
(18/70: ils /prãd/ replaces ils /prEn/) and the 4PR (16/70: nous /prãdõ/
replaces nous /prənõ/).
In the following analysis, we focused on the errors in these three target
forms where two specific conditions were fulfilled: (1) an existent stem was
used; (2) an appropriate inflection was used.
When the target form was the 6PR, the two conditions were fulfilled in
110/200 errors (55%), while the infinitif stem was used in only 11 errors
(10%). The inflection of the 4PR was generally correctly chosen: 112/152 or
73.6% of errors were marked with the correct inflection /õ/. However, the use
of existent stems for this target form was rare. Existent stems were used in
only 31 errors, and an infinitif stem was used in only 4 errors.
As for the 3FUT, the rate of correct inflections when an existent stem
was used was 64% (31/48). The two conditions were fulfilled in 17% of
incorrect futur simple forms (31/179). We examined the stems that were used
to build the futur simple. As can be seen in Table 13, the infinitif stem was
most often used to replace a futur simple stem. Therefore, the errors in the
futur simple represent the most salient cases of the use of the infinitif stem to
build other forms.
Table 13. Errors in stem choice in the futur simple classified as a function of
substitute stems
Substitute stem Number Example
infinitif 19 fEra for fəra
Présent 1&2pl., imparfait 8 byvra for bwara
Présent 3pl. 2 bwavra for bwara
Présent sg. 2 dwara for dəvra
Total 31
Further evidence that the infinitif is used to build the futur simple lies
beyond the scope of our error analysis. We therefore reanalyzed the mean
accuracy rates of futur simple stem selection in two groups of verbs: first, the
verbs which share the same stem in the infinitif and the futur simple; second,
the verbs in which these two stems are different. In our task, this distinction
contrasted the verbs prendre, craindre, boire and traduire that use a single
stem for the futur simple and the infinitif with the verbs faire and devoir
which use two different stems. The mean rates of correct futur simple stem
choice in these two groups of verbs are presented in Table 14.
The acquisition of French verbal tenses by Russian adult learners 195
Table 14. Correct futur simple stem choice in verbs which use one or two stems for
the futur simple and the infinitif (mean rate and standard deviation)
We compared the rates of correct futur simple stem choice for these two
verb types. Among the LP group productions, we observed a tendentially
higher rate of correct stem choice for the verbs that use the same stem for the
infinitif and the futur simple (t=2.006, p=0.072). This trend did not, however,
affect the productions of the HP learners group (t=1.414, p=0.188).
To check whether or not the greater accuracy of stem choice on verbs
that use the same stem for the infinitif and the futur simple in the productions
of the LP group might be due to their higher frequency, we compared the
mean frequencies of the stems of the two groups of verbs in the NSF corpus
which corresponded to the input of the LP learners. The mean frequency of
the futur simple stem of the verbs which use two different stems is 39.38
(SD=25.50) per million words. This rate is not significantly different from
the mean frequency of the futur simple stem of the verbs which use a single
for the infinitif and the futur simple (M=27.21, SD=21.11) (t=0.882,
p=0.398). This suggests that frequency does not explain the greater accuracy
with verbs that use one stem for the futur simple and the infinitif on the part
of the learners in the LP group. In such conditions, it is clear that the infinitif
stem is used to build the forms of the futur simple.
More generally, the implementation of Bybee’s criteria (Bybee 1985)
for the definition of the base forms of the verb paradigm leads us to conclude
that the stems of the présent 3sg. and the stem of the infinitif play this role for
Russian learners of French L2.
These results support the linguistic model proposed by Bonami & Boyé
(2003) as well as Bybee’s behavioral model, since these both predict the base
form status of the présent singulier. It should be noted that similar results
were obtained by Thomas (2005) who studied the acquisition of French
verbal morphology by Swedish learners. It is possible to imagine that the
base form status of the présent singulier and the infinitif might be explained
by the structure of the French language, given that the same results have been
obtained for learners with a different L1.
196 Evgueniya Sergeeva & Jean-Pierre Chevrot
5. Conclusion
statistical patterns of use of verbal forms in the textbook or in the more global
input.
Second, using Bybee’s criteria, we tried to determine whether the French
verb really has a base form for Russian learners of French L2.
• To identify what the base forms of the French verb might be during L2
acquisition, we analyzed verbal forms in the light of the idea that a base
form should be semantically non-marked, frequent, acquired at an early
stage of language development and used to build the paradigm (Bybee
1985). First, we found that three forms were distinguished by a high rate
of accuracy - the 3PR, INF and the PP - and that this was observed in
both levels of language learner. The forms in question were not equally
likely to be used to build other forms. Indeed, the 3PR was often used to
replace other forms directly. The INF was not used to replace forms as a
whole, but operated as a stem for building the futur simple forms with a
non-zero inflection. The PP does not meet all the criteria of a base form:
it is semantically marked and is not often used to replace other forms.
These findings are interesting bearing in mind that, in French, the choice
of the stem can be crucial in expressing differences in person, tense or mood
when there is a zero inflection. We have shown that stem choice is influenced
by at least two factors that are different from the notions of verbal tense,
person and mood, i.e. token frequency and the base form status of the présent
singulier and the infinitif. In other words, even if L2 learners have already
acquired conceptual differences between different verbal categories, the
processing of the corresponding surface forms can be facilitated or inhibited
by the token frequency of the targets, by the frequency of other competing
stems of the same verb and by the fact that the target stem is different or
identical to one of the base form stems.
References
Tabak W.; Schreuder R.; Baayen R.H. (2005). The processing of regular and
irregular verbs. Proceedings of the Interdisciplinary Workshop on the
Identification and Representation of Verb Features and Verb Classes.
Saarbrücken, 121-126.
Thomas A. (2005). Formes ambiguës en –E: input et catégories sémantiques.
Acquisition et production de la morphologie flexionnelle. Actes du
Festival de la morphologie. PERLES: petites études romanes de Lund.
Extra seriem 20, 35-52.
Ullman M.T. (2001). The neural basis of lexicon and grammar in first and
second language: the declarative/procedural model. Bilingualism:
Language and Cognition. 4 (1): 105-122.
From tense and aspect to modality:
The acquisition of future, conditional and subjunctive
morphology in L2 French. A preliminary study 1
Martin HOWARD
University College, Cork
1. Introduction
1
I gratefully acknowledge the support of the College of Arts, Celtic Studies and
Social Sciences, University College, Cork in facilitating research for this paper
through a research grant.
time contexts in particular before the past time forms emerge. In the case of
her instructed adult learners in the US, Bardovi-Harlig (2000) reports that the
first of these forms to emerge is the simple past followed by the past
progressive, and then the present perfect, and finally the pluperfect. Housen
(2002) offers similar evidence in the case of his investigation of child L2
learners of English in an immersion school setting in Belgium. Even when
such morphology is present in learner language, however, its usage is
characterised by considerable under- and over-use, whereby some linguistic
contexts appear to (dis)favour use of a particular form, giving rise to
considerable variation at work in the learner’s aspectuo-temporal system –
see Bayley (1996) who examines such aspectuo-temporal variation from a
variationist perspective.
Such gradual emergence and subsequent variation underlying the
acquisition of aspectuo-temporal morphology has been observed in a range of
L2s. In the case of L2 French, which is of specific interest to the study to be
presented here, the sequence of acquisition reflects that of the simple present
form in L2 English, following which some past participles begin to be used,
before being used with an auxiliary form to constitute the passé composé.
While the imparfait emerges even later, the learner may already have
produced some forms of être (to be) and avoir (to have); – for detailed
analyses of such development in L2 French, see Dietrich et al. (1995), Harley
(1992), Noyau (1991), Schlyter (1990), Véronique (1987). Similarly, the
plus-que-parfait, equivalent to the pluperfect, is very late to emerge - see
Howard (2005a). In the case of the various forms used to express past time,
the linguistic variation previously referred to in the case of L2 English has
also given rise to a range of fruitful studies of L2 French, whereby a number
of patterns of variation have been observed to be at work as the learner gains
increasing control over the use of the various past time forms in real time -
see Howard (2004, 2005c), Kihlstedt (1998) and Labeau (2005) for an
investigation of the factors believed to constrain the learner’s use of
aspectuo-temporal morphology in relation to grammatical and inherent
lexical aspect, as well as discourse grounding.
While the emphasis has been very evidently on the expression of past
time, more recent studies have looked beyond the past to the expression of
futurity, as well as modality and irrealis involving the conditional and the
subjunctive. Although studies are few, in the case of L2 English, K. Bardovi-
Harlig’s more recent work has focused, for example, on the gradual
emergence of ‘will’ and ‘going to’ and their subsequent competition with the
simple present and present progressive forms in future time contexts—see
Bardovi-Harlig (2002, 2004a, 2004b, 2005). In the case of L2 French,
Nadasdi et al.’s (2003) work focuses on the variable use of the French
equivalents of such forms from the alternative perspective of the
development of their Canadian immersion learners’ sociolinguistic
The acquisition of morphology in L2 French 203
Although she does not include the progressive form in this acquisition order,
other work by Giacalone Ramat (1995) indicates that this form is late
acquired in L2 Italian, following the imperfect. In the case of future time
reference in L2 Italian, Wiberg (2002) provides comparative studies of
bilingual Swedish adolescents, L2 learners at both an advanced and pre-
advanced stage, and native speakers. Her findings, in particular, demonstrate
that while the learners use the inflected future form, their use of this form
alternates with the simple present as a function of verb type in terms of
inherent lexical aspect, as well as the discursive context, whereby learners
have a tendency to present future events as habitual actions in the present.
In the case of L2 Spanish, results of existing studies also reflect aspects
of the hierarchical order observed for L2 Italian. For instance, with regard to
the expression of past time specifically, various studies by Roger Andersen
have outlined the following acquisitional sequence, as summarized by
Andersen (1991): firstly, the unmarked base form, similar to the third person
singular present, is used in past time contexts, after which an opposition
2
It should be noted that the morphological forms concerned are not necessarily
directly comparable across the Romance languages—although they are in many
regards similar, forms such as the subjunctive and the conditional are seen to
play a more important role in Spanish and Italian.
204 Martin Howard
between this form and the perfective préterito appears. It is only at a later
stage that the imperfecto is contrasted with the préterito. Salaberry (1999)
offers similar results, finding that the préterito is used as a default form until
the imperfecto emerges at a much later stage. Apart from the past time forms,
the subjunctive has also given rise to a number of fruitful studies in L2
Spanish—for example, Terrell et al. (1987) note the minimal use of this form
in their learners’ speech, giving rise to a high frequency of error, which
contrasts with their far higher level of metalinguistic awareness surrounding
its usage, as tested on a formal grammar test. Use of the subjunctive in L2
Spanish has also been the focus of study in relation to the impact of various
pedagogical methods. For example, in an extensive study of the relationship
between syntactic and morphological competence, Collentine (1995, 1998)
examines how development of the learner’s syntactic knowledge might be a
pre-requisite to more appropriate use of the subjunctive. His findings are
ambiguous, however, insofar as he fails to find an absolute advantage for his
learner-informants who received instruction in relation to both the
morphological and syntactic use of the subjunctive, as opposed to simply
learning the subjunctive.
In conclusion, the acquisition patterns put forward for L2 English and
the various Romance languages are very insightful, reflecting the very
extensive investigation that tense-aspect-mood has undergone. However, in
the case of L2 French in particular, the focus on individual forms, namely the
subjunctive or the conditional or the expression of futurity, without
comparing how the acquisition of such forms relates both to each other, and
to previously acquired forms, is regrettable. As Bardovi-Harlig (1999: 373-4)
writes, “[W]ork in the tense-aspect system in interlanguage has generally
avoided the third element 3 in what is referred to as the “TMA” or “tense-
mood-aspect system” […] At the level of description, we need to explore the
hypothesized sequences for the spreading of morphology, for which only the
earliest stages have been documented.” Such comments are particularly true
given that we know that the acquisition of an individual form does not occur
without impinging on use of previously acquired forms in view of the
morpho-syntactic variation and competition which we know underlie the
learner’s use of morphological devices. Indeed, in the case of L2 French it is
only by a process of indirect inference that we can surmise that the
morphological expression of the conceptual entities of futurity, modality and
irrealis is late acquired compared to past time morphology; for example,
Bartning (1997) observes that such morphological forms are generally
missing even in advanced learner language, so that their acquisition remains
to be documented. Based on a more extensive advanced learner corpus,
3
Although ambiguous here, the third element refers to modality in Bardovi-
Harlig’s paper.
The acquisition of morphology in L2 French 205
Bartning (2008) indicates that such forms are only observed in the very
advanced stages - stages D and E in Bartning & Schlyter’s (2004) 5-stage
developmental sequence for Swedish learners of French. However, it remains
unclear how, with the various forms concerned here, their acquisition is
interrelated, so that their acquisition may be inter-dependent as well as being
subject to considerable variation in terms of their contexts of usage. Given
such a lacuna in the literature, the study presented here aims to provide a
comparative investigation of their acquisition, based on their analysis in a
database of advanced L2 French which has already been the subject of a
range of studies concerning the expression of past time—see Howard (2002a,
2002b, 2005b).
Before outlining the study undertaken, we will in the following, provide
some background detail on the use of the forms concerned in target language
French.
The morphological choice that such forms pose to the speaker has been the
subject of extensive prescriptive characterisation, whereby prescriptive
grammars have attempted to attribute distinct semantic values to each form—
see, for example, Confais (1995), Fleischman (1982), Gougenheim (1971),
Sundell (1991), Vet (1993). Although not always in accord, such prescriptive
approaches typically assign to the periphrastic future and futurate present
semantic values of certainty, immediacy, and speaker engagement, whereby
the occurrence of the event in the future is believed to be conceptualized as
certain, proximal and of particular relevance to the speaker. In contrast, the
inflected future prescriptively assumes opposing values whereby its potential
4
A future anterior form also exists, but is less the focus of analysis here.
206 Martin Howard
5
Groupe aixois de recherche syntaxique.
6
Although the phonetic distinction between /e/ and /є/ is prescribed, the
distinction is increasingly found to only occur in very formal speech.
The acquisition of morphology in L2 French 207
(7). This differs from a less realistic condition, the outcome of which is
perceived as improbable, in which case the conditional is prescribed in the
apodosis, along with the imparfait in the protasis, as in (8).
(6) ‘si’ + Present + Present:
si vous échouez à l’examen une deuxième fois,
if you fail-PRES at the exam a second time,
vous n’avez pas le droit de le repasser une troisième fois
you NEG have-PRES NEG the right to it resit a third time
‘if you fail the exam a second time, you don’t have the right to sit it a third time’
(7) ‘si’ + Present + Inflected Future:
si vous m’appelez à l’avance, je viendrai
if you me call-PRES at the-advance, I come-INFL FUT
vous rencontrer à la gare
you meet at the station
‘if you call me in advance, I will come to meet you at the railway station’
(8) ‘si’ + Imparfait + Conditional:
si vous achetiez une voiture, je ne devrais pas
if you buy-IMP a car, I NEG have to-COND NEG
vous rencontrer à la gare chaque fois
you meet at the station each time
‘if you bought a car, I wouldn’t have to meet you at the railway station every
time’
7
The past subjunctive consists of a composed form, much like the passé
composé, and is prescribed in contexts where the occurrence of the event is
anterior or simultaneous to the subjunctive-inducing verb.
The acquisition of morphology in L2 French 209
meaning that the subjunctive form has lost a lot of ground in terms of the
extent of its usage, sometimes not occurring at all following matrix verbs
which supposedly require its usage. Indeed, S. Poplack finds that the
subjunctive only occurs with any regularity with 4 lexical verbs, namely
‘aller’ (to go) (87%), ‘faire’ (to do / make) (86%), ‘avoir’ (to have) (66%),
and ‘être’ (to be) (65%). Furthermore, this subjunctive usage co-occurs with
a very limited range of subjunctive-selecting matrices, namely ‘falloir’
(must), which along with ‘vouloir’ (to want) and ‘aimer’ (to like) “together
account for nearly ¾ of all the subjunctive governors in the corpus” (Poplack
2001: 417). In a further Canadian study, Laurier (1989) also documents the
loss in use of the subjunctive among francophone speakers in Ontario, so that
its usage is primarily found to follow ‘falloir’ (must) in the main clause,
similar to Poplack’s findings. In contrast, findings for metropolitan French
point to more robust usage of the subjunctive, demonstrating little social
variation. For example, in the case of her findings from the GARS project on
spoken French, Blanche-Benveniste (1990: 197) writes, “d’après les corpus
que nous avons rassemblés au GARS, le mode subjonctif n’est pas du tout
moribond ni même en passe de l’être” (based on the corpora that we collected
within the GARS, the subjunctive mood is far from being lost or even on its
to being lost). Based on his investigation of the Orléans corpus, Sand (2003)
similarly observes the highly productive usage of the subjunctive form.
Having presented an overview of previous work on their usage in native
speaker discourse, we will in the following detail the study undertaken which
aims to document the acquisition and use of the various forms concerned in
advanced learner language.
3. Study
4. Data analysis
4.1. Futurity
All morphological forms used for future time reference were extracted from
the data, giving rise to 116 tokens. Future time reference was determined
based on H. Reichenbach’s model, whereby future events involve a relation
of posteriority between the speech time and the event time. In cases where
such a relation was ambiguous, other contextual clues were used to determine
the future time reference of the event, including the use of temporal adverbs
and adverbials, as well as the temporal frame provided by the informant’s
interlocutor.
In the case of the subjunctive, the data extraction was determined by the
presence of subjunctive-inducing verbs in the matrix clause as well as
subordinating conjunctions prototypically requiring the subjunctive in native
speaker discourse. The data did not demonstrate use of superlative adjectives
following which the subjunctive might be expected. In all, 100 subjunctive-
inducing contexts were identified, and subsequently coded for (non-)use of
the subjunctive, as well as for verb (ir)regularity whereby the subjunctive
8
Apart from such linguistic variables, the data have also been the source of
analysis in a project on the acquisition of sociolinguistic variation. For an
overview, see Howard (2006b).
212 Martin Howard
form is only salient on irregular forms across all persons, as well as on 1st and
2nd person plural forms of regular verbs.
Similar to the previous markers, all 215 tokens identified were coded for
their morphological form and the actual lexical verb concerned. In cases
where the learners produced a main clause dependent on a si-clause, the
morphological marking of both verbs concerned was also coded with a view
to investigating the type of tense-concordancing that the learners might
engage in across clause types.
Having outlined the study undertaken, we now present the results for the
three groups of learners as previously detailed, namely the classroom learners
of group 1, the study abroad learners of group 2, and the learners in group 3,
who had received a further year’s instruction compared to the learners in the
other groups.
5. Results
A first observation concerns the high frequency of use of the present as well
as the conditional in group 1, which contrasts with this group’s non-use of the
periphrastic future and of the conditional anterior, as well as its minimal
realisation of salient subjunctive forms. Occurrence of the inflected future is
a little more robust at 28%.
In the case of group 2, we can similarly observe this group’s very
frequent use of the conditional form, which constitutes the most frequent
form of those studied. Its relative ease of use is also reflected in the
conditional anterior whose rate of use reaches 44%, albeit consisting of just 4
tokens out of a possible 9 where the form could have occurred. While the raw
number of tokens is very low, it is nonetheless notable that neither of the
other two groups produce this form. Concerning the other forms investigated,
we also note that their usage is more robust compared to group 1, particularly
in the case of the inflected and periphrastic future forms, at 38% and 31%
respectively, while the subjunctive remains the least frequent form at just
16%.
By comparing the findings for groups 1 and 2 with those of group 3, we
can see how the classroom instruction that the learners in group 3 have
experienced may have led to differential use of the forms investigated
compared to the naturalistic exposure experienced by the learners in group 2.
In this regard, we observe a number of similarities and differences. Firstly, as
in the case of the other groups, use of the conditional is by far the most
frequent form, which contrasts with the more minimal use of the subjunctive
and the periphrastic future form. In the case of the latter form in particular,
such a tendency is more similar to group 1. However, in contrast to group 1,
group 3 demonstrates much more frequent use of the inflected future form,
exceeding the previously noted frequent use by group 2. In so doing, group 3
also relies much less heavily on the present for future time reference, as is the
case with group 2, but in contrast to group 1.
214 Martin Howard
The study presented aimed to illuminate the acquisition and use of some
lesser investigated TMA forms in L2 French. By adopting a global focus as
opposed to focusing on a specific form or a single functional context such as
the expression of futurity or conditionality, the study reveals some interesting
The acquisition of morphology in L2 French 215
case of the expression of futurity, the learners have recourse to the present
form which already exists in both their interlanguage and the target language.
In contrast, in marking a hypothetical situation, the conditional is not used
variably with a present form in target language French. However, it is
important to note that the learners’ use of the present in future contexts
cannot necessarily be seen to reflect its status as a futurate present in the
target language; rather the learners may simply be using it as a default marker
of futurity, much as the present constitutes a default marker of past time, in
particular in imperfective contexts, in the acquisition of past time reference—
for discussion see Howard (2005c). Indeed, this form also appears in
conditional contexts when the learners fail to use conditional forms—while
some other forms do also appear in place of the conditional, the present is by
far the most frequent.
Apart from the present and the inflected future forms, the learners rely
much less on the periphrastic future, except in the case of our study abroad
learners. The limited use of the periphrastic future is reflected in the
subjunctive and the conditional anterior, which, taken together, are situated at
the opposite end of the spectrum to the more frequent forms of the
conditional, (default) futurate present, and inflected future 9.
As a whole, figure 1 allows for an interesting overview comparison of
the relative frequency of the forms concerned in relation to other existing
acquisition continua for both L2 French and other L2s. A comparison with
such continua presented in studies like Bartning & Schlyter (2004) for L2
French, as well as Giacalone Ramat (1992) for L2 Italian points to some
differences and similarities with our findings 10. In particular, the very late
acquisition of the subjunctive reported in both studies is reflected in our
learners’ minimal use of the form in contexts where its usage would be
required in target language French. While our learners produce a range of
9
While the conditional anterior does demonstrate a rate of use of 44% in group 2,
we situate this form at the end of the continuum presented in order to reflect the
difficulty that its usage poses to the learners in the other groups which do not
use this form at all.
10
Unlike the acquisition order presented by A. Giacalone Ramat, we do not
include our previous findings for the past tenses, since these are not the focus of
analysis here. Suffice to say that our learners, irrespective of their grouping
here, demonstrated very high levels of accuracy of use on the passé composé
and the imparfait, ranging from 75%-92%, as reported on by Howard (2005b).
While our previous work has illuminated the variation at work in use of such
past time forms, such high levels of accuracy point to their relative ease of use
compared to the forms under analysis here. In sum, such relative stability on
past time marking may be a prerequisite for enhanced development on the forms
presented here, reflecting A. Giacalone Ramat’s findings for L2 Italian, as well
as those of Schlyter (1990) for her Swedish naturalistic learners of French.
The acquisition of morphology in L2 French 217
input exposure may favour the formal variant, namely the inflected form,
whereby classroom learners tend to overuse formal variants at the expense of
informal ones. Such an interpretation is abundantly supported in the literature
on the acquisition of sociolinguistic variation in the L2—see, for example,
Dewaele & Mougeon (2003), Howard (2006b), Rehner et al. (2002).
The positive impact of the naturalistic exposure experienced by our
study abroad learners in group 2 is further reflected in their use of the
conditional anterior which the other groups do not use at all, while they also
demonstrate a higher rate of accuracy on tense-concordancing in conditional
clauses in general, as well as on marking (non-)realistic hypotheses in
particular. However, in contrast with their relatively frequent use of the
periphrastic future and the conditional anterior, the study abroad group does
not demonstrate any advantage in relation to the other forms investigated.
Rather, it is our classroom learners in group 3 who demonstrate an equally
important effect for the classroom instruction they have received: they use
the forms at a level similar to or slightly more frequently than their study
abroad counterparts. This is true for example in their use of the conditional,
as well as the subjunctive, albeit used relatively infrequently overall. The
positive effect for their programme of classroom instruction is also reflected
by the learners in group 3 insofar as they demonstrate considerable progress
compared to their less advanced counterparts in group 1. They rely much less
on the present to mark futurity, while also demonstrating a superior level of
accuracy on tense-concordancing in conditional clauses.
In conclusion, while the effect for learning context, in the case of the
forms concerned here is extremely relative, the results entail important
implications for language pedagogy aimed at advanced learners in the foreign
language classroom. In particular, by illuminating the relative frequency of
use of the various forms, the results indicate areas where the advanced
learner may need to be presented with particular opportunities to productively
use the forms in order to facilitate their more frequent occurrence in the
learners' real time communication than is currently the case, as reflected in
the results presented here.
References