|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
ACQUISITION OF TENSE-ASPECT MORPHOLOGY BY ENGLISH LEARNERS OF
FRENCH AND CHINESE
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
To My Parents
I wish to thank my advisor, Dr. Theresa A. Antes, for her consistent
encouragement, support, and help throughout my doctoral studies and particularly for her
assistance with this dissertation. Thanks go to other committee members Dr. Jean
Casagrande and Dr. Diane Boxer, especially to Dr. Joaquim Camps and Dr. Chauncey
Chu, for their expert input and suggestions on theoretical as well as methodological
issues. The guidance from all of these professors and many others at the University of
Florida has helped put me on the right track to be a mature and independent scholar.
I am also indebted to the following good friends and colleagues: Sophie
Ganachaud, Barbara Petrosky, David Petrosky, Bin Li, Jinping Zhu and many others who
have helped with the design and analysis of the activities.
Finally, my gratitude goes to my husband Xuan Meng for his love and
encouragement. It is also important that I acknowledge my son Raymond Meng who has
been a sweet and cooperative boy throughout the project.
TABLE OF CONTENTS
A C K N O W L E D G M E N T S ................................................................................................. iii
LIST OF TABLES ............................................. ........... .............. .. vii
A B ST R A C T .......... ..... ...................................................................................... x
1 INTRODUCTION ............... ................. ........... ................. ... .... 1
2 TH EORETICAL ISSU ES ................................................. ............................... 5
T ense and A aspect ............................................................... .. ... ......... 5
T e n se .................................................. 6
Aspect ............. .. .............................. .... .........................7
Interaction betw een tense and aspect ...................................... ............... 10
Tense and Aspect in French and Chinese ........................................ ................11
F rench ............................................................... ................... ........ 11
C hinese ............................................................... .... ..... ... ...... 14
T ense in C hinese ............................................ ............ .. ........... ..14
A spect in C hinese ................... .................................... .... .... .. ........ ... 18
Summary of Grammatical Tense and Aspect..........................................................22
L ex ical A sp ect ............................ ........................................................... ............... 2 4
Between Lexical Aspect and Grammatical Aspect .......................................... 27
3 ACQUISITIONAL STUDIES.......................... ........................................... 30
Aspect Hypothesis and Related Studies .......................................... ............... 30
Major Studies on the Aspect Hypothesis .......................... .......... .. ..............33
D iscourse H ypothesis ..... ...... ...... ...... .. ............ .... ............... 36
Studies on the Acquisition of French as a Second Language..................... ........ 37
U ntutored L earners ................................................................... ....................37
Tutored Learners .................................. .. .. ...... .......... .....41
Im m version learners .................. ................. ........ ...... .... .................. ..41
Beginning and intermediate foreign language learners.............................43
Advanced foreign language learners: ........................... .................. 46
Studies on the Acquisition of Chinese...................... ...... ........ .................. 48
L1 Acquisition of Chinese Temporality ............................................................49
L2 Acquisition of Chinese Temporality ............................................................52
Distributional Bias Hypothesis and Prototype Hypothesis.............. ............ 56
4 M E T H O D O L O G Y ............................................................................ ................... 6 1
P a rtic ip a n ts ........................................................................................................... 6 1
French and Chinese Programs ........................................................62
French and Chinese Textbooks........................................................ ............... 63
T a sk s .........................................................................................................6 6
D ata C collection P procedures ............................................................. .....................73
D ata E n co din g ................................................................7 3
5 R E S U L T S .............................................................................7 8
F ren ch D ata .......................................................7 8
C o m p o sitio n s ................................................................................................. 7 9
Cloze test ................................... ................... ....... ...............87
Summary of the results on French data .......................................................... 92
C h in e se D ata ......................................................................................................... 9 2
C om p reh en sion T ask ...........................................................................................92
C o m p o sitio n s ................................................................................................. 9 5
Written Editing Task ...............................................104
Summary of the results of the Chinese data ...............................................107
6 D ISC U SSIO N ...................................................... 109
R research Question O ne........................................................ 109
Answer to Question One .............................................. ...............109
Comparisons with Previous Studies of French ............................................... 109
R research Q u estion T w o ...................................................................................... 111
A nsw er to Question Tw o ................................................................... .. .. ....111
Comparisons with Previous Studies of Chinese ................................................ 112
Research Question Three .............................................................. .. .............113
R research Im plications........... ............................................................ ............121
P edagogical Im plication s .................................................................................... 12 5
F ren ch .............................................................................................. ....... 12 5
C hinese ...................................................................................................126
Lim stations of the Present Study ..................................................... ..... ...........130
7 C O N C L U SIO N ...................................................................................... 132
A LANGUAGE BACKGROUND INFORMATION .................................................134
B FRENCH COMPREHENSION TASK ....................... ......... ............... 136
C FRENCH CLOZE TEST ........................... ............................. ...............138
D CHINESE COMPREHENSION TASK.... ......... ....... ............ 140
E CH IN E SE CL O ZE TE ST ......... ................. ........................................................142
F C O M PO SIT IO N T O PIC S ............................................................... ....................144
G OPERA TION A L TE STS................................................. ............................... 145
L IST O F R E FE R E N C E S ......... ................. ...................................... ..........................146
B IO G R A PH ICA L SK ETCH ......... ................. ..........................................................152
LIST OF TABLES
2.1 Classification of aspectual opposition (Comrie, 1985, p25)..................................9
2.2 Aspectual opposition in English, French and Chinese .......................................23
2.3 Semantic features of the four verb classes: ................................ ............... .....24
4.1 Inform ation on the participants ........................................... ......................... 62
4.2 Semantic features used in the verb classification................... .................76
5.1 Accuracy of the French comprehension over the four aspectual classes
(percentage) .............. ................. ............................... ...................... ................78
5.2 Verb forms used in the three French compositions and their distribution over
four aspectual classes (percentages (raw frequencies)...........................................79
5.3 Distribution of individual verbs used in Imp and PC over the four aspectual
classes in the three French compositions (percentage (raw frequencies)................80
5.4 Accuracy rate for Imp vs. PC in the three French compositions............................81
5.5 Accuracy of Imp and PC over the four lexical aspectual classes in the three
French compositions (percentage (appropriate uses/relevant contexts)) .................82
5.6 Distribution of Imp and PC by composition ..........................................................83
5.7 Distribution of PC and Imp over the four aspectual classes in French
composition 1 (percentage (raw frequencies)) ......................................................84
5.8 Distribution of PC and Imp over the four aspectual classes in French
composition 2 (percentage (raw frequencies)) ......................................................84
5.9 Distribution of PC and Imp over four aspectual classes in French composition 3
(percentage (raw frequencies)).......................................................... ............... 84
5.10 Accuracy of use of PC and Imp over the four aspectual classes in French
composition 1 (appropriate use/relevant contexts = percentage).............................84
5.11 Accuracy of use of PC and Imp over the four aspectual classes in French
composition 2 (appropriate use/relevant contexts = percentage).............................85
5.12 Accuracy of use of PC and Imp over the four aspectual classes in French
composition 3 (appropriate use/relevant contexts = percentage).............................85
5.13 Use of Imparfait with different verb classes by individual participants .................85
5.14 Use of Passe Compose with different verb classes by individual participants ........86
5.15 Use of Imparfait with different verb classes by participants who produced all
fo u r c la sse s ...............................................................................................................8 7
5.16 Use ofPasse Compose with different verb classes by participants who produced
all four classes .........................................................................87
5.17 Distribution of the answers provided by native speakers over the four aspectual
classes in the French cloze test (percentage (raw frequencies)) ..............................88
5.18 Distribution of verb forms over the four aspectual classes in the French cloze
test in learners' production (percentage (raw frequencies)) .....................................88
5.19 Accuracy of PC and Imp used in relevant contexts over the four aspectual
classes in the French cloze test (percentage (appropriate uses/relevant contexts))..90
5.20 Accuracy of the answers in the Chinese comprehension task ..............................93
5.21 Accuracy of the answers in the Chinese comprehension task over the four
aspectual classes (percentage) ........._...._.................... ........... 93
5.22 Particles used in the three Chinese compositions................... .................96
5.23 Aspect markers used and their distribution over the four aspectual classes in the
three Chinese compositions (percentage (raw frequencies))...............................97
5.24 Aspect markers used by composition in the Chinese compositions (percentage
(raw frequency)) .................................................. .......... .. ......... 99
5.25 Accuracy of the aspect markers used by composition (percentage (appropriate
uses/relevant contexts)) ............................. ............................. ............... .99
5.26 Distribution of the particles over the four aspectual classes in Chinese
composition 1 (percentage (raw frequencies))................................................... 100
5.27 Distribution of the particles over the four aspectual classes in Chinese
com position 2 (percentage (raw frequencies))...................................................... 100
5.28 Distribution of the particles over the four aspectual classes in Chinese
com position 3 (percentage (raw frequencies))...................................................... 100
5.29 Accuracy of the particles over the four aspectual classes in Chinese composition
1 (percentage (appropriate use/ relevant contexts)) ..............................................101
5.30 Accuracy of the particles over the four aspectual classes in Chinese composition
2 (percentage (appropriate use/relevant contexts)) ....................................... ...... 101
5.31 Accuracy of the particles over the four aspectual classes in composition 3
(percentage (appropriate use/relevant contexts)) ............................................. 101
5.32 Distribution of the aspect markers in relevant contexts over the five different
verb structures in Chinese compositions (percentage (raw number)) ....................101
5.33 Accuracy of the particles over the four lexical classes in the Chinese
compositions (percentage (correct use/ relevant contexts)) ....................................102
5.34 Accuracy of the particles over the five verbal structures in the Chinese
compositions (percentage (raw number)).......................................... .............103
5.35 Answers supplied by two native speakers in the Chinese written editing task and
their distribution over four aspectual classes (percentage (raw number)).............. 105
5.36 Distribution of all the particles over the four aspectual classes in learners'
answers in the Chinese written editing task (percentage (raw frequencie...............05
5.37 Accuracy of the particles in relevant contexts in the Chinese written editing task... 106
5.38 Accuracy of the particles over four lexical classes (percentage (correct uses/
relevant contexts)) ........ .......... ... .................................... .. ........................ 107
Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
ACQUISITION OF TENSE-ASPECT MORPHOLOGY BY ENGLISH LEARNERS OF
FRENCH AND CHINESE
Chair: Theresa A. Antes
Major Department: Romance Languages and Literatures
The present dissertation investigates the acquisition processes involved in the
learning of French and Chinese tense-aspect morphology by English speakers. The
objective is to test the Prototype Hypothesis, which states that, in general, learners first
acquire the prototypical members of a grammatical category and gradually extend their
knowledge to the non-prototypical members. The Prototype Hypothesis treats tense-
aspect morphology as a prototype category, and it states that the prototypical features of
the perfective aspect (or simple past tense) are [+ result], [+ punctual] and [+ telic]
whereas the prototypical features of the imperfective aspect are [- result], [- punctual] and
[- telic]. The Prototype Hypothesis is developed from the Aspect Hypothesis and explains
the relationship between tense-aspect morphology and the inherent lexical aspect of the
The dissertation starts with an explicit explanation on how the tense-aspect systems
work in both Chinese and French. Following the theoretical explanation is a review of the
previous studies done on the acquisition of tense-aspect morphology, with a special focus
on the acquisition process in learning French and Chinese. The present dissertation is
based on an empirical study where learners performed a series of comprehension and
production activities including one comprehension task, three compositions, one cloze
test in French and one written editing task in Chinese. The results from the French data as
well as from the Chinese data showed support for the Prototype Hypothesis. Both the
distribution and the accuracy in the use of tense-aspect morphology in both languages
were influenced by the inherent lexical aspect of the predicates. The effect was more
evident in French than in Chinese. More traces of the influence of lexical aspect were
found in the compositions than in the cloze test among the learners of French whereas
more evidence was found in the written editing task than in the compositions among the
learners of Chinese.
The differences in the results between Chinese and French can be explained by the
differences in the tense-aspect systems of the two languages. Lexical aspect plays a more
important role in the tense-aspect system in French than in Chinese.
The present dissertation contributes to the research on the acquisition of tense-
aspect morphology in two ways: first, it provides much-needed empirical evidence on the
influence of lexical aspect in the acquisition of Chinese; and second, it gives a contrastive
analysis of the acquisition processes of an Indo-European language and a non-Indo-
Tense and aspect has long been the focus of language pedagogy, but the
investigation on the acquisition of tense-aspect morphology in second language
acquisition began to draw specific attention in the mid-1980s.
In the study of the acquisition of temporal systems, three concepts are crucial:
tense, grammatical aspect and lexical aspect. Tense locates a situation on a time line
(Comrie, 1985). Grammatical aspect concerns the internal temporal constituency of a
situation (Comrie, 1976). Lexical aspect refers to the characteristics inherent in the
lexical items that describe the situation. Predicates can be classified into four lexical
aspectual classes according to their inherent meaning: states, activities, accomplishments
and achievements. The Aspect Hypothesis, proposed by Andersen (1991, 1994) is an
influential hypothesis on the acquisition of tense-aspect morphology in first and second
language acquisition. It points out the relationship among these three concepts in the
acquisition process and states that" 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, p. 133). More precisely, it predicts that the acquisition of imperfective aspect will
start with states and gradually spread to activities and accomplishments before finally
reaching achievements. The spread of perfective aspect is predicted to move in the
opposite direction: from achievements to accomplishments then to activities and finally
The Aspect Hypothesis has received ample support from many Indo-European
languages, including English, French, Italian, Portuguese, Russian, and Spanish, as well
as from non-European languages such as Japanese and Chinese. While there are many
studies done on Indo-European languages in the acquisition of tense-aspect, the research
on non-Indo-European languages is sparse. Especially in second language acquisition, the
number of studies is very limited.
The Aspect Hypothesis is based on observations from empirical studies on the
learning processes of tense-aspect morphology. To explain these observations, Shirai and
Andersen (1995) turns to the Prototype Hypothesis, which states that there are
prototypical and non-prototypical members of a grammatical category and that the
learners first acquire the prototypical members of a grammatical category before
gradually extending their knowledge to the non-prototypical members. In the acquisition
of tense-aspect morphology, for instance, the acquisition of imperfective aspect, states
share the most prototypical features among the four aspectual classes with imperfective
aspect. This explains why the Aspect Hypothesis predicts that the imperfective marking
starts with states before moving to other verb classes.
The objective of the present dissertation is to see how tense-aspect morphology is
learned by English-speaking learners of French and Chinese. It is designed to test the
Prototype Hypothesis with data from French and Chinese. In other words, the study aims
to answer the following questions:
1. Is the use of imparfait and passe compose by intermediate-level learners of French
as an L2 influenced by the inherent semantic aspects of verbal predicates?
2. Is the use of the grammatical aspect markers (-le, -guo and zai) by intermediate-
level learners of Chinese as an L2 influenced by the inherent semantic aspects of
3. How does the acquisition of grammatical aspect in the two languages differ in
terms of the influence of the aspectual classes?
Both French and Chinese have a perfective versus imperfective distinction in their
grammatical aspectual system. The present study will focus on two past tenses in French:
imparfait imperfectivee) and passe compose perfectivee) and four aspect markers in
Chinese: perfective marker -le, experiential marker -guo, durative marker -zhe and
progressive marker zai. There are overlaps of functions among the aspect markers of the
same grammatical category imperfectivee or perfective) between the two languages, but
difference is dominant when we compare the two. For instance, in French, tense and
aspect are fused together in one morphological marker, while in Chinese, the aspect
markers are independent of tense. In view of the entirely different tense-aspect systems in
the two languages, it is interesting to see if one factor of the system, lexical aspect, works
the same way in the acquisition of these forms.
There have been several studies done on French that tested the Aspect Hypothesis
using quantitative analysis and verb classification to see the influence of lexical aspect.
Generally speaking, in all the levels tested, from beginners to advanced learners, the
influence of lexical aspect has been found. A few studies that touched on the issue of
aspectual classes in Chinese had mixed results. These studies on Chinese were not
designed to test the predictions of the Aspect Hypothesis, but the results in them can be
used as evidence to either support or reject the Aspect Hypothesis. The present study will
provide additional evidence to the interaction between aspectual classes and tense-aspect
morphology in Chinese.
The present study differs from all others in that it investigates the acquisition
process of two distinct languages: French, of the Romance Language family, and
Chinese, of the Sino-Tibetan Language family. It is the first study that compares an Indo-
European language with a non-Indo-European language. The comparison of the two
languages within the same research design will offer insights on the following issues: 1.
Are the claims in the Aspect Hypothesis universal? 2. How does the acquisition process
of tense-aspect differ from one Indo-European language to one non-Indo-European
The present study is organized as follows: Chapter 2 introduces various theoretical
issues: what are tense and aspect? How are tense and aspect represented in French and
Chinese respectively? Chapter 3 reviews previous studies on the acquisition of tense-
aspect morphology in first and second language acquisition with a special focus on the
studies conducted on the acquisition of French and Chinese. Chapter 4 presents the
methodology of the present study in detail. It introduces the participants, the data
elicitation methods, and the data analysis used. Chapter 5 summarizes the results studies.
The explanation and discussion of the results, along with pedagogical and research
implications and limitations of the present study, are also presented in Chapter 6 and
Chapter 7 draws the conclusion.
Tense and Aspect
All natural languages have ways of expressing the concept of time; what
differentiates them is whether they assign more weight to grammaticalization or to
lexicalization. Lexicalization refers to lexical means of expression, for example
adverbials such as a week later, lastyear, tomorrow, etc. Grammaticalization refers to
ways of establishing time that require obligatory expression and morphological
boundness (Comrie, 1985). English past/non-past opposition is a clear case of a
grammaticalized way to locate time, because it is obligatory even when the context
(especially the adverbials) makes the time reference clear; the bound morpheme is the
past tense marker (-ed for regular verbs) on the verbs. The set of adverbials is infinite, as
people keep creating new phrases, while the set of grammaticalized forms is very limited.
It should be specified that grammaticalization is manifested on the verbs, meaning
that in most languages that have tense, the tense is indicated on the verb, either by the
verb morphology or by grammatical words adjacent to the verb, as with auxiliaries.
The present study is concerned with the grammaticalized forms used to express the
concept of time in French and Chinese. By contrasting the two systems, we hope to
answer some questions concerning second language acquisition, specifically, the
acquisition of time reference in Chinese and French by English speakers.
The grammaticalized means of expressing time in languages include tense and
aspect. The former refers to the way one locates a situation1 on a time line (present, past
or future) and the latter concerns the internal temporal constituency of a situation. Tense
is deictic since it relates a situation to a reference point, and aspect is non-deictic.
Tense can be divided into three categories: absolute tenses, relative tenses and
combined absolute-relative tenses. Absolute tenses take the present moment as their
deictic center while relative tenses take a reference point (which could be in the present,
past or future) as their deictic center, i.e. the time of the relative tenses depends on the
Present, past and future tenses belong to absolute tense. If we use S to represent the
moment of speech and E for event time, these tenses can be represented using the
following formulae (Comrie, 1985):
Present tense: E simultaneous to S
Past tense: E before S
Future tense: E after S
It should be noted that there is a core meaning associated with individual tenses
while each of them has other meanings in specific languages. For instance, the core
meaning of the English past tense is to represent past time reference, but it has several
other secondary, yet very important, meanings, e.g. counterfactual present, as in ifI were
you, and polite requests, as in I just wanted to ask ifyou could help me i/it.... What
1 Following Comrie (1985), situation here is used as a general cover-term, i.e., a situation may be a state, an
event or a process.
determines if a grammaticalization can be called a certain tense or not is if it fits the core
meaning of the definition. The 'core' meaning is the prototypical meaning.
All the non-finite verbs in English and French can be said to have relative tense
because their tense is dependent upon the tense of the finite verbs occurring in the same
context. In the following examples:
1). Walking around the neighborhood, I always see Peter.
2). Walking around the neighborhood, I always saw Peter.
The phrase "walking around the neighborhood' has two different interpretations as
far as tense is concerned, with the former being in the present and the latter in the past.
Using R as the reference point, the formulae for relative tenses are (Comrie, 1985):
Relative present: E simultaneous to R
Relative past: E before R
Relative future: E after R
English and French pluperfect would be examples of a combined absolute-relative
3). English: He had left when she arrived.
4). French: Il 1titparti quand elle est arrive.
Here, the situation 'his leaving' is located before the reference point 'her arrival'
which is in turn located before the moment of speech. So 'her arrival' is in absolute tense
while 'his leaving' is in relative tense.
The formula for pluperfect is: E before R (before S).
As mentioned above, aspect indicates the internal temporal constituency of a
situation (Comrie, 1976). "Aspect" here refers only to grammatical aspect, and does not
include lexical aspect, which will be explained later. In other words, it refers to those
aspectual distinctions that are grammaticalized in languages. The difference between the
following two sentences is one of aspect:
5). He read a book.
6). He was reading a book.
The tense of both sentences is past, yet the first situation is viewed as a single
whole perfectivee) and the second as having a process with no specific beginning or
ending imperfectivee). We can conclude from the example that aspect is 'viewpoint', i.e.,
it is how people view the situation, so aspect is said to be subjective because the same
situation can be described with different aspects without any contradictions.
The basic aspectual opposition is between perfective and imperfective, as shown in
the example above. Perfective aspect looks at the situation from outside without an
attempt to divide the situation up into various individual phases. Imperfective aspect
views the situation from within and is concerned with the internal temporal structure of a
situation (Comrie, 1976).
Imperfective aspect is further divided into several subcategories in some languages,
as shown in table 2.1. Imperfective aspect includes habitual aspect and continuous aspect
which is composed of progressive aspect and its counterpart, non-progressive aspect. The
identifying feature of habitual aspect is that it describes a situation, which is characteristic
of an extended period of time, so that the situation is viewed as a characteristic feature of
the whole period. The structure used to in English is a form of habitual aspect. The
general definition of progressiveness is the combination of progressive meaning and
nonstative meaning. English Progressive copulaplus verb ing manifests the progressive
Table 2.1 Classification of aspectual opposition (Comrie, 1985, p25)
"Perfect" is also traditionally treated as an aspectual category, although it does not
look at the internal temporal constituency of a situation, but, rather, relates a state to a
preceding situation. For example, the present perfect indicates the continuing present
relevance of a past situation. The past perfect2 expresses a relation between a past state
and an even earlier situation; and the future perfect expresses a relation between a future
state and a situation prior to it. The preceding situation can happen in the past, present, or
Aspectual categories, like tense and all other grammatical categories, have a core
(prototypical) meaning and other secondary meanings. English Progressive serves as a
good example to illustrate this point. The core meaning of the Progressive is to describe
an ongoing nonstative situation. For example, John is doing his homework. But the
English Progressive goes far beyond this core meaning and also has the following
functions (Comrie, 1976):
1. To express a more temporary state as opposed to a more permanent state expressed
by its non-progressive counterpart. Compare the following sentences:
7). I'm living at 321 SW 76th street.
2 Pluperfect is treated in grammatical tradition as a tense, but it is actually, according to linguistic
classification, a combination of tense (past) and aspect (perfect). Many so-called tenses in grammatical
tradition are a combination of both tense and aspect.
8). I live at 321 SW 76th street.
2. To add an emotive effect that is lacking in the non-progressive, as illustrated in the
9). I've had two whiskies and already I'm seeing pink elephants. (Comrie, 1976, p.
37) The progressive here gives the hint that I'm only imagining things.
Different aspects can be combined to describe a situation. For example, the English
Habitual Past and the Progressive can be used together to form sentences like he used to
be singing.... The Perfect can also be combined with progressive in English to render the
structure like I've been wondering....
Interaction between tense and aspect
In many languages, we find that grammaticalized aspectual distinctions are made in
the past tense but not in other tenses, suggesting that it may be a general characteristic of
human languages to resort to greater aspectual differentiation in the past than in other
tenses. English Habitual aspect only exists in the past and the French
perfective/imperfective distinction also only exists in the past. In the Narrative Present
(historical present) in French, where the present tense is used to refer to a past situation,
the perfective/imperfective distinction is lost. Since the present tense is essentially used
to describe, rather than to narrate, it is essentially imperfective, not perfective. While
there is present perfective in some languages, it exists primarily to express a perfective
non-past which has a future meaning. For example, in Russian, the Perfective non-Past is
primarily a future tense. Though perfective non-past does not exist in English as a
grammatical category, the point can still be illustrated with examples in English. The
sentence: he comes here on its own, normally has a habitual meaning, since if the
reference were to an action going at the present moment, it would have to be he is coming
here, i.e., Progressive. However, in the context where he comes here does not have
present time reference, then perfective meaning is a possible interpretation, as in a
subordinate clause of time, e.g., when he comes here, I'll tell him. Here the verb comes
refers to a future action (Comrie, 1976).
In Arabic, the opposition Imperfective/Perfective incorporates both aspect and
(relative) tense (Comrie, 1976). The Perfective indicates both perfective meaning and
relative past time reference, while the Imperfective indicates everything else (either
imperfective meaning or relative non-past tense). The same thing occurs in Chinese,
where the Perfective marker -le indicates at the same time aspect and relative tense (see
the section entitled Perfective marker -le).
Tense and Aspect in French and Chinese
Because aspectual distinctions exist in particular in the past tense in the languages
in question, past time reference will be the focus of the following section, in which an
overview of the tense and aspect system is outlined in French and Chinese, with
comparisons made to English.
In French, both tense and aspect are grammaticalized. Every finite verb in the
language is conjugated for tense and aspect. The opposition perfective/imperfective exists
only in the past tense. Two main tenses are used to express past in spoken French: passe
compose (compound past; expressed hereforth as PC) and imparfait (imperfect; expressed
hereforth as Imp). PC has perfective meaning while Imp has imperfective meaning. Both
tense and aspect are represented on the verb form and it is impossible to distinguish
which part of the verb indicates tense and which indicates aspect. The fusion of the
morphological markers of aspect and other categories in French (and other Romance
languages) explains why forms which are differentiated aspectually, such as PC and Imp,
are traditionally referred to as different tenses, rather than aspectual forms of the same
Here is one example of the PC form for the verb parler (to speak):
j'aiparlk (I spoke; I have spoken; I did speak)
tu asparlk (you-sing. spoke; you have spoken; you did speak)
il/elle aparlk (s/he spoke, s/he has spoken; s/he did speak)
nous avonsparlk (we spoke; we have spoken; we did speak)
vous avezparlk (you-sing. orpl. spoke, you have spoken; you did speak)
ils ontparlk (they spoke; they have spoken; they did speak)
We can see from the example that the PC is composed of two parts: an auxiliary
avoir meaning "to have" (or &tre meaning "to be") and the past participle. The only part
that is conjugated is the auxiliary in the present tense.
PC was called perfect in some traditional grammars, but it covers both perfect and
nonperfect meanings. See the following example:
10). I1 afini ses devoirs. (He has finished his homework or He finished his homework.)
In the development of French, there has been a gradual reduction of the presentness
of the perfect (PC), which finally became purely past (Comrie, 1985). The PC can be
used as the English Present Perfect to relate a past state to a present situation to indicate
the continuing present relevance of a past situation (as in I have spoken). It can also be
used as English Simple Past with perfective meaning (as in I spoke).
From the core meaning of past tense and perfective aspect respectively, we can
conclude that the basic meaning for PC is to locate a situation before the moment of
speech and to view the situation as a whole. This basic meaning gives rise to some basic
temporal functions of PC:
1. To express a situation that ended in the past.
11). J'ai &crit une lettre. (I wrote a letter.)
2. To express a situation that has present relevance from a situation that ended in the
12). Je vous ai, dos la publication, reserve un exemplaire. (I have, from the time of
the publication, reserved a copy for you.)
3. To express a situation completed within a time scope that has not ended.
13). Aujourd'hui, j 'ai achetd un livre. (I bought a book today.)
4. To indicate a situation which no longer exists in the present but was habitual during
a well-defined period of time.
14). Pendant trois semaines, il n 'a mange que dupain sec. (For three weeks, all he
ate was dry bread.)
5. After si (if), to replace the future anterieur (anterior future).
15). Si vous avezfini dans une heure, je vous donnerai cinq francs. (If you have
finished in an hour, I'll give you five francs.)
Here is an example of the Imp of the same verb that was used to present PC: parler
je parlais (I used to speak; I was speaking; I spoke)
tuparlais (you sing. used to speak; you were speaking; you spoke)
il/elle parlais (s/he used to speak; s/he was speaking; s/he spoke)
nousparlions (we used to speak; we were speaking; we spoke)
vousparliez (you-pl. or sing. used to speak; you were speaking; you spoke)
ilsparlaient (they used to speak; they were speaking; they spoke)
The Imp can be translated into English habitual past, past progressive and simple
past with imperfective meaning. The main temporal functions of Imp include:
1. To describe people, things or facts as they were in the past;
16). Savez-vous qui &tait son mari? (Do you know who her husband was?)
2. To express an habitual action in the past;
17). Ils me donnaient un cadeau tous les ans pour mon anniversaire. (They used to
give me a gift every year for my birthday.)
3. To express circumstances that accompany a principle action in the past; in other
words, to describe the background of a situation;
18). I1 neigeait quandje suis arrive a Paris. (It was snowing when I arrived in
4. To describe situations already started and continued in the past;
19). Mon frdre &tudiait l'anglais depuis deux ans. (My brother had been studying
English for two years.)
The difference between PC and Imp is not one of tense, as both express past tense
reference, but one of aspect with the first perfective and the second imperfective. Imp is
often used to describe the background for events in PC to take place.
Tense in Chinese
Most Chinese linguists argue that Chinese is a tenseless language but that it is rich
in aspect (Smith, 1991; Erbaugh, 1992; Li & Shirai, 2000). Chinese has a
perfective/imperfective aspectual distinction which exists in present, past and in future
contexts; in addition, Chinese has a perfect aspect. The perfective marker is -le while the
imperfective markers include zai and -zhe; -guo and le are perfect markers. Among these
markers, the interpretation of LE3 is the most controversial, and maybe the most difficult
part of Chinese grammar for second language learners. There has been discussion
concerning LE to determine whether there is one LE or two LEs. The present study
adopts the two-LE-position (Smith, 1991; Erbaugh, 1992; Chu, 1998) and states that
there is a verb-final perfective -le and a sentence-final le that indicates change-of-state
with a focus on the perfective -le.
According to Comrie (1985), tense is a grammaticalized way of locating a situation
in time. There are not only absolute tenses which take the moment of speech as a
reference point, but also relative tenses that take another situation as reference point. If
we think about tenses only from the angle of absolute tenses, Chinese could be
considered a language without tense, since there is no specific morpheme for present,
past or future tenses. Consider the following examples:
20). Wo meitian qu chaochi.
I everyday go supermarket
'I go to the supermarket everyday.'
21). Wo zuotian qu chaochi.
I yesterday go supermarket
'I went to the supermarket yesterday.'
22). Wo mingtian qu chaochi.
I tomorrow go supermarket
'I will go to the supermarket tomorrow.'
As we can see in these examples, the time reference changes from the present, to
the past and then to the future from sentences 20 to 22, but the verb qu 'to go' remains
3 The present paper follows the convention and marks the verb-final perfective marker as -le and the
sentence-final perfect marker as le while LE is used to represent both collectively.
unchanged. However, there are two problems with the examples given above. First,
sentence number 21 sounds unnatural to a native ear, as if it were not finished; a native
speaker would expect more information to come. Second, it is impossible to treat time
reference with isolated sentences in Chinese, just as it is impossible to treat relative
tenses with only subordinate clauses in English.
The verb-final particle -le is traditionally considered an aspect particle with
perfective meaning, which can be used in present, past or future contexts (Li and
Thompson, 1981; Smith, 1991; Chu, 1998; Xiao, 2001). In the previous examples,
sentence number 21 is in a past context, which is the most common context where the
perfective -le is found, but this particle is frequently associated with present and future
contexts as well:
23). Women xia-le ke jiu qu kan dianying. (future context)
We down-le class then go see movie.
'We'll go see a movie when we get out of the class.'
24). Ta meitian chi-le wanfan hou sanbu yi xiaoshi. (present context)
S/he everyday eat-le dinner after walk one hour.
'S/he takes a walk for an hour everyday after dinner.'
Since the perfective -le can be used in present, past and future contexts, it cannot
be treated as an absolute past tense marker (besides its function as a perfective marker),
but it is reasonable to treat it as a relative past tense marker. By definition, relative past
tense locates a situation (E) before a reference point (R) while the absolute tense of the
reference point is determined by context (Comrie, 1985). This definition does not exclude
the present moment or the moment of speech as reference time, actually, the absolute
tense of the reference point can be any of the three: past, present or future.
Chang (1986) proposes that the verb-final perfective -le has an anteriority-marking
function. When there is a series of situations that happen, the particle -le is used at the
end of the first verb to indicate that the second situation doesn't happen until after the
first one. This anteriority-marking function can be reasonably interpreted as a relative
past tense marking, because it fits the definition exactly: to indicate one situation that
happens prior to another situation, which is used also as the reference point.
Ross (1995) claims specifically that the perfective -le should be treated both as a
past tense marker and as a perfective aspect marker. Based on the framework of Bull
(1971 cited in Ross, 1995), the characterization of tense is the anchoring of a situation to
an axis of orientation. The perfective -le indicates that the situation to which it is suffixed
is bound by and anterior to the situation presented in the following VP. The axis of
orientation (or the reference point) is the following VP. In short, the perfective -le signals
anteriority with respect to an axis of orientation. Basically, the past tense that Ross talks
about is relative past tense in Comrie's terminology, since it is with respect to the
The present study will adopt Ross' analysis and will call the verbal -le both a
perfective marker and a relative past tense marker. As with the French PC, the perfective
marker -le covers both perfect and perfective meaning in English. As Ross points out, a
characterization of verbal -le purely in terms of aspect obscures its function of relative
anteriority. What's more, the anteriority-marking function explains the non-use of -le.
Before examining this non-use of-le, however, let us first give a complete literature
review of how perfective -le is treated in Chinese linguistics.
Aspect in Chinese
Perfective -le. As mentioned above, most Chinese linguists agree that -le should be
considered a perfective aspect marker. What differentiates different scholars is the exact
definition of 'perfectivity' they give to -le. It is generally agreed that -le is obligatory,
but an agreement has not been reached as the circumstances under which it is obligator.
In other words, -le's grammatical functions are still under discussion. It is no easy task to
summarize where to use -le because, with different contexts, it could be both
ungrammatical and obligatory within the same sentence. In other words, the
grammaticalityof -le goes beyond the sentence-level.
Li and Thompson (1981, p185) characterize the function of-le as 'bounded' and
give four criteria to determine the boundedness of a situation:
1. By being a quantified event
2. By being a definite or specific event
3. By being inherently bounded because of the meaning of the verb
4. By being the first event in a sequence
The boundedness characterization explains -le's function better than previous
claims where -le is said to signal completion (Chao, 1968, cited in Chu & Chang, 1987).
Comrie (1985) states that the perfective aspect does not have completion in its meaning.
Yet the problem remains that boundedness cannot account for the many exceptions where
the situation is bounded but -le is not used. For example:
25). Zuotian wo qu shudian mai-le yi ben shu.
Yesterday I go bookstore buy-le one [MeasureWord] book.
'I went to the bookstore yesterday and bought a book.'
The situation "I went to the bookstore" is a bounded event since it satisfies two of
the conditions above: it is a definite event and the first in a sequence. However, -le is not
needed with the verb 'go', although it is used with the verb 'mai'.
In an attempt to explain the non-use of the perfective marker -le, Chu (1998)
proposes to examine the question from a discourse perspective. This discourse
perspective does not mean grounding; actually, Chu points out that grounding does not
work here. Andreason (1981, cited in Chu 1998) has claimed that -le has the function of
marking foreground in narrative discourse, but many foregrounded clauses occur without
-le. This point can be illustrated with the same example just mentioned. In example 25,
the situation "I went to the bookstore" is in the foreground, but no -le is needed. The
solutions advanced by Chu are peak-marking function and anteriority-marking function.
We have discussed anteriority-marking above and called it the function of a relative
past tense marker. This tense-marking function overlaps with one of the functions of -le
as a perfective aspect marker, which comes as no surprise given that there is a natural
association between perfective aspect and past tense. The anteriority-marking function
derives from the notion of perfectivity but it should be understood in a discourse context
where it signals that the second situation does not happen until after the first situation
(where -le is suffixed) ends. This explains why the marker -le is not needed for the
second situation, but required for the first.
The peak-marking function is also a discourse function which, similar to the
anteriority-marking function, derives from the notion of perfectivity. It views a whole
series of situations as a single one and suffixes -le to the peak situation. This explains
why in a series of clauses, only one of them is suffixed with -le, while the others are not.
Special attention should also be paid to occasions when -le should not occur.
Besides the discourse functions of -le that explain its non-use in discourse, some specific
verbs often reject -le. They are the 'say' verbs (shuo 'say', gaosu 'tell', tongzhi 'inform',
etc.) and verbs with a classic flavor (i.e., verbs from old Chinese). -Le is not combined
with the 'say' verbs either in direct or indirect speech. Perfective -le is not used here
because the focus is the content of what is said. However, when the focus shifts from the
content of speech to whether or not the person did the telling, -le could be used with the
'say' words. This characteristic proves that the perfective -le marks the most salient
message in a discourse. The reason why verbs with classical flavor do not take -le is
explained by the late emergence of the affix (Mei Tsulin, 1981 cited in Chu 1998).
Perfect le. It should be noted that there is another le in Chinese that is
homographic and homophonic to the perfective -le. This other le occurs in sentence-final
position while the perfective -le occurs in verb-final position. The function of le is to
signal a change-of-state. For example,
26). Ta gao le.
S/he tall le.
'S/he is tall now.' (meaning s/he was not tall before)
This change-of-state could be simply a realization on the part of the speaker,
though, and not necessarily a change in the objective situation. Consider the following
27). Xia yu le.
Down rain le.
'It is raining.' (It has started raining.)
This sentence can be used not only when it has just begun to rain, but also when the
speaker has just discovered that it is raining.
Li, Thompson and Thompson (1979) propose that we consider le a perfect aspect
marker even though le is sentence-final and not verb-final. They claim that the basic
communicative function of le is to signal a Current Relevant State (CRS), that is, le
denotes that a state of affairs has special current relevance to some particular reference
time. Furthermore, they adopt a 'cluster' view of the universality of functional categories,
namely, that functional categories such as aspect should be treated as having a core
meaning and other secondary meanings specific to individual languages. As the core
meaning of the perfect aspect, which is the CRS, is the same as that of Chinese le, they
argue that le should be treated as a perfect aspect marker. Using or not using sentence-
final le changes the meaning, but the sentences by themselves are always grammatically
correct with and without le. Sentence-final le and verb-final le can occur in the same
Experiential perfect marker -guo. Following Comrie (1985), the present study calls
-guo the experiential perfect marker because it indicates that a situation has been
experienced at least once in the past. Compare the following sentences to contrast -le and
28). Ni chi-le yuchi ma?
You eat-le shark's fin?
'Have you eaten shark's fin?' (Did you eat shark's fin?)
29). Ni chi-guo yuchi ma?
You eat-guo shark's fin?
'Have you ever eaten shark's fin?'
Imperfective markers. Both (zheng) zai and -zhe indicate ongoing situations, so
they can be categorized as "continuous", under the imperfective aspect. The differences
between the two lie in three areas:
1. Zai is used before the verb while -zhe is a suffix.
2. Zai indicates progressive situations while -zhe can occur with progressive and non-
progressive situations (i.e. non-states and states). "Progressive" means the
combination of continuous meaning and nonstative meaning. In different phases of
a progressive situation, the situation changes (or progresses), while in a continuous
situation with non-progressive feature, the different phases represent exactly the
same state. Compare the following sentences:
30). Ta xihuan ting-zhe yinyue xie zuoye.
S/he like listen-zhe music write homework
'S/he likes to do his homework while listening to music.'
31). Ta zhengzai ting yinyue, turn you ren qiao men.
S/he zhengzai listen music suddenly have people knock door
'Someone knocked at the door when s/he was listening the music.'
3. Zai indicates that the clause with the prefix is temporal in nature while -zhe
indicates that the clause with the suffix is a treated as a manner adverbial. Zai is
used more often to answer a 'when' question, and -zhe a 'how' question. Sentences
30 and 31 can be used again to illustrate this third point.
Summary of Grammatical Tense and Aspect
To summarize, as table 2.2 shows, in the past tense, French and English both have
absolute and relative past tenses, but the relative tense is not grammaticialized. Chinese,
however, only has relative past tense. Aspect plays an important role in further dividing
the past tense category. In French and English, tense and aspect are fused into one form
and every form has both the functions of tense and aspect. In Chinese, the perfective
aspect marker -le also has a relative past function and all the other markers have only
Table 2.2 Aspectual opposition in English, French and Chinese
English French Chinese
Perfective passe compose -le
Imperfective imparfait *
Progressive be + V-ing zai
Perfect have + present le, -guo
participle of Verb
French has a perfective/imperfective distinction; this opposition is denoted by PC
and Imp respectively. In the imperfective, &tre en train de is the progressive form
lexicalizedd not grammaticalized way to express progressive) in French, but it is not often
used; instead, the Imp is often used to translate the English progressive form. For
example, David was writing will be rendered in French David &crivait.
Chinese has a perfective/imperfective distinction as well as a perfect aspect form.
The perfective marker is -le, the imperfective markers are zai and -zhe, and the perfect
markers are -guo and le.
In English, there are only progressive and perfect aspects, otherwise, there is just
the Simple Past form, with no further distinction of aspect. Habitual aspect used to exists
in the past, but it is optional because the simple past can replace it.
Both the French and Chinese perfective aspect incorporates the perfective as well
as the perfect meaning in English. The French Imp covers the progressive meaning but
goes beyond it because Imp can also be used for stative situations. Both English and
Chinese have progressive and perfect aspects while French does not.
4 All the aspect markers in Chinese have other functions besides tense or aspect functions; the present paper
is only concerned with their functions within the tense-aspect system
When talking about aspect, it is important not to limit the discussion to
grammatical aspect only. Lexical aspect is equally important in the aspectual system.
Lexical aspect, also called situation aspect or Aktionsart, refers to the characteristics
inherent in the lexical items that describe the situation. As Andersen and Shirai (1996)
noted, the inherent lexical aspect referred to here includes only the morphosyntactic
aspectual information in a sentence, such as verbs and their arguments, not temporal
Vandler (1967) classified verbs in English into four categories: states, activities,
accomplishments and achievements. This classification is based on three semantic
features: telicity (with or without an endpoint), dynamicity, and durativity (Shirai &
Andersen, 1995, p744).
Table 2.3 Semantic features of the four verb classes:
Punctual Telic Dynamic
Accomplishment -+ +
Achievement + + +
With the help of Table 2.3, the four verb classes (Shirai & Andersen, 1995) are
defined as follows:
* States are verbs that have no dynamics, and continue without additional effort or
energy being applied. Examples of states are to be, to believe, to think, etc.
* Activities are verbs that are dynamic in nature, with no clear endpoint, and the
phases in the internal structure of an activity are homogeneous. Examples of
activities are to run, to sing, to dance, etc.
* Accomplishments differ from activities in that they do have a clear endpoint.
Examples of accomplishments are to build a house, to run a mile, to sing a song,
* Achievements are punctual, telic and dynamic, they differ from accomplishments in
that they happen instantaneously. Examples of achievements are to realize, to
enter, to break, etc.
It can be seen from the examples above that even though the classification is based
on verbs, their arguments are not to be ignored. The same verb can belong to different
aspects depending on the complements it takes, thus "to sing" is an activity, but "to sing a
song" is an accomplishment and "to sing songs" is an activity again. The present paper
follows Andersen and Shirai in calling the verb classification inherent lexical aspect.
Different from what Smith (1991) calls "situation type," inherent lexical aspect does not
change with adverbials or with tense-aspect morphology; it is concerned only with the
verb and its arguments within the situation defined by the whole sentence.
Vandler's four-way classification is widely accepted and used as a theoretical basis
for many linguistic studies. The classification is "a linguistic universal" and "a cognitive
universal" (Andersen & Shirai, 1996, p. 532). It has been applied to Indo-European as
well as non-Indo-European languages (Jacobsen, 1982; Foley & Van Valin, 1984; Weist
et al., 1984; Holisky, 1981 cited in Andersen & Shirai, 1996; Li & Bowerman, 1998; and
Smith, 1991; among others). Following previous studies, the present paper adopts the
four-way classification to categorize verbs in French and Chinese.
Grammatical aspect is an overt grammatical category, while lexical aspect is a
covert grammatical category (Erbaugh, 1992). In other words, one can determine from
the form of the verb which grammatical category it belongs to, but the lexical aspect of a
verb can only be learned individually and there are no affixes to indicate it. We have
mentioned, in our discussion of tense and aspect in general, that the grammatical aspect
in different languages has the same basic meaning (core meaning), but is not necessarily
identical because every grammatical category in a language has its own specific, even
idiosyncratic, features. Inherent lexical aspect works the same way. Attention should be
paid to the following two points: 1. One category, for example, achievements, does not
necessarily have the same set of verbs in different languages. The translation of a verb
into another language does not mean a match in lexical aspect between the two verbs. 2.
The rules concerning verb complements are not the same in different languages either.
Thus, although two verbs in different languages might have the same semantic meaning,
one could belong to different inherent lexical aspectual classes while the other can only
belong to one.
The relationship between grammatical aspect and lexical aspect has been
extensively investigated in linguistic theory, in studies of first language (L1) acquisition
as well as second language (L2) acquisition, both in terms of language universals and
contrastive analysis. It is evident that grammatical aspect and lexical aspect interact and
that there is a close relationship between the two in terms of combinations between them
(Comrie, 1976; Shirai & Li, 2000). Natural associations have been observed, for
example, between perfective aspect on the one hand and achievements and
accomplishments on the other; between imperfective aspect and states; and between
progressive aspect and activities. This comes as no surprise, since there exist considerable
overlaps between grammatical aspects and lexical aspects. For example, the
characteristics for perfective aspect overlap with those for achievements, namely
punctuality and termination (endpoint). It should be pointed out here that all grammatical
aspects can combine with verbs in all four lexical aspects, it is just that the natural
associations mentioned above are more frequent (or more prototypical).
Between Lexical Aspect and Grammatical Aspect
The verbs in both French and Chinese can be classified according to Vendler's
four-way classification; however, before going into the relationship between lexical
aspect and grammatical aspect in the two languages, it is important to note some special
features of the Chinese verb system. The category 'verb' in Chinese is quite different
from that in English or French. First of all, the verbs do not conjugate for tense, person,
voice or mood; second, the distinction between verbs and adjectives is not so clear. There
exists a category in Chinese called "adjective predicate" (Hsu, 1998), which resembles
more the category of adjectives in English. For example:
32) Ta hen gaoxing.
S/he very happy.
S/he is (was) very happy.
In this example, there is no 'verb' and the adverb hen (very) directly modifies the
'adjective' gaoxing (happy). In the category 'adjective predicate', the adjective takes the
role of the verb "to be." Chu (1983) claims that adjective predicate should be called state
verbs whereas other verbs be called action verbs. The present study follows Chu and
classifies these adjective predicates as states.
Another special category within the Chinese verb system is the Resultative
Compound. They are composed of a verb and a Resultative Complement to represent the
meaning 'finish doing something'. Example are given below:
33). zuo wan
do finish 'finish doing'
V + Resultative Complement
33). zhuang fan
knock over 'knock over'
V + Resultative Complement
As stated in Chu (1976 cited in Xiao, 2001), Chinese action verbs alone only
indicate active attempts rather than attainment of goals, it is with the help of perfective
marker -le and resultative complements that action verbs reach a goal. Resultative
compounds form a logical whole, and it is not practical to separate the verb from its
resultative complement. Chinese often uses resultative compounds to describe events that
English specifies with accomplishments and achievements. Unlike English
accomplishments, however, Chinese resultative compounds cannot be marked with the
progressive marker. Following Li and Bowerman (1998), the present study treats
resultative compounds as a subclass of achievements.
As mentioned in previous sections, there is a perfective/imperfective aspectual
opposition in the French past tense. The perfective past tense is passe compose
(compound past, PC) and the imperfective past tense is imparfait (imperfect, Imp). Both
PC and Imp can be combined with all four types of verbs. However, the associations
become weaker when we go from the combination between PC and achievements and
accomplishments to the combination between PC and activity and even weaker in the
combination between PC and states. Imp works in the opposite direction: the associations
between Imp and states are the strongest and become weaker with activities,
accomplishments and then with achievements. Actually, the associations between PC and
states, as well as those between Imp and achievements are so weak, that they both require
a shifted interpretation (Smith, 1991, p. 255). The former requires an inchoative reading
and the latter requires a repetitive reading. Comrie (1975) also talks about these unusual
combinations and his idea is in line with that of Smith. Comrie claims that these
combinations limit the range of interpretation of a given verb.
Chinese also has a perfective/imperfective distinction. The perfective marker is
-le while the imperfective markers include zai and -zhe; in addition, there are two perfect
aspect markers: le and -guo.5 All the aspect markers can be combined with all the four
verb classes. As in French, however, unusual combinations, such as a perfective marker
with states or an imperfective marker with achievements, require shifted readings.
It should be noted that the definition of "situation type" in Smith (1991) is different
from the one adopted by the present paper. He claims that the situation type changes from
states to inchoatives (which are of the accomplishment or achievement situation types in
Smith, 1991) when states are combined with a perfective aspect marker. In our view,
however, the inherent lexical aspect does not change; what changes is the situation type
of the whole sentence. That is why we clearly indicate that all aspect markers can occur
with all four verb types. But we should be aware that not all aspect markers can occur
with all the verbs in every verb class. For example, in English, states do not usually
combine with progressive aspect. When we say that progressive aspect can be combined
with all the four verb classes, it means that it is possible for some of the states to be
marked by progressive aspect, not that every state predicate can be used in progressive. It
is grammatically correct to say he is being silly, but incorrect to say *he is knowing the
fact (Comrie, 1976).
5 It is essential to note that there are two LEs, but only verb-final -le is the focus of the present study.
Aspect Hypothesis and Related Studies
In the L1 and L2 acquisitonal studies on tense-aspect morphology, one observation
is constantly reported: learners seem to restrict their tense-aspect marking to certain verb
classes. For example, learners use simple past in English only with punctual situations
(i.e. achievements) and use progressive aspect only with activities. Several hypotheses
have been proposed to describe what is going on, including Bickerton's (1984) Punctual-
Non-Punctual Distinction (PNPD) and State-Process distinction, which essentially claim
that children mark those distinctions instead of tense. Bloom et al. (1980 cited in
Andersen & Shirai, 1996) also claim that aspect is learned before tense in the sense that
children mark lexical aspect first, not tense. Weist et al. (1984) do not agree with Bloom
et al. and call Bloom's claim the Defective Tense Hypothesis. Using studies on the
acquisition of Polish as evidence, Weist et al. claim that children mark both tense and
aspect. Overall, however,the most influential hypothesis is the Aspect Hypothesis
(Andersen, 1991, 1993; Andersen & Shirai 1994, 1996). The Aspect Hypothesis
(Andersen & Shirai, 1994, p. 133) states that "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."
More specifically, the Aspect Hypothesis claims that perfective past inflections are
predominantly attached to achievements and accomplishments in the early stages, and
that imperfective past marking, which emerges later, is used predominantly with state-
activities initially. As Andersen and Shirai (1996) caution us, the Aspect Hypothesis
makes an observational description of inherent lexical aspect and grammatical tense-
aspect marking pairings during acquisition without attempting to offer a cognitive
explanation behind the learning process.
Andersen and Shirai (1996, p. 533) summarize the descriptive claims of the Aspect
Hypothesis as follows:
1. Children first use past marking (e.g. English) or perfective marking (e.g. Chinese,
Spanish, etc.) on achievements and accomplishments, eventually extending its use
to activities and states.
2. In languages that encode the perfective-imperfective distinction, imperfective past
appears later than perfective past, and imperfective past marking begins with states
and activities, then extending to accomplishments and achievements.
3. In languages that have progressive aspect, progressive marking begins with
activities, and later extends to accomplishments or achievements.
4. Progressive marking is not incorrectly overextended to states.
As can be seen above, the Aspect Hypothesis is not concerned with the question of
whether tense or aspect is acquired earlier. It simply states the relationship observed in
language acquisition between inherent lexical aspect and tense-aspect morphology. In
other words, it does not explain what drives the learners to create a certain interlanguage,
it just describes the patterns found in the interlanguage. Tense-aspect morphology is
viewed as the final product of the learners, not the starting point.
The Aspect Hypothesis also predicts that there are eight developmental stages
associated with the distribution of tense-aspect morphology across verb classes
(Andersen, 1991, p. 315): at Stagel, learners mark neither past tense nor aspect; at Stage
2, the use of the perfective aspect (or simple past tense) is encoded in punctual verbs
only; at Stage 3, prototypical state predicates appear in imperfective forms; at Stage 4, the
perfective spreads to accomplishments, while the imperfective spreads to activities: all
verbs are now marked by inherent aspect in past tense; at Stage 5, the use of verbal
morphology begins to overlap within each verb type: telic verbs can now be marked by
imperfective or perfective aspect; at Stage 6, activities can be used with perfective or
imperfective aspect; at Stage 7, punctual events can be marked by either imperfective or
perfective aspect; Stage 8 constitutes the end of the sequence: states can now be encoded
in perfective aspect.
The Aspect Hypothesis points out a new direction in the study of acquisition of
tense-aspect morphology. Many studies in L2 acquisition have tested it on different
languages and given it ample support. Early studies focused on a small number of target
languages, namely Spanish, English and French (Salaberry, 1998, 1999; Kumpf, 1984;
Robison, 1990; Bardovi-Harlig, 1992, 1998; Kaplan, 1987; Bardovi-Harlig & Bergstrom,
1996 among others), then later expanded to languages such as Catalan, Dutch, Italian,
Japanese, Portuguese and Russian (Comajoan, 1998 and Housen, 1994; Giacalone Ramat,
1995, 1997; Shirai, 1995 and Leary, 1999). Learners involved in the studies include
tutored and untutored learners, foreign language learners and second language learners.
The data elicitation methods range from personal and impersonal narratives, to written
narratives, to silent film retells, to cloze passages.
All four claims seem to be supported with L2 acquisition studies. The association
of perfective past (e.g. English simple past, Spanish preterite) with achievements and
accomplishments is by far the most robustly attested stage in the distribution of verbal
morphology in the interlanguage system. The developmental sequences described by
Andersen are also observed in cross-sectional studies. In languages where the opposition
between perfective/imperfective exists (French, Italian and Spanish), imperfective
appears to be acquired later than perfective and states seem to be strongly associated with
imperfective (Harly, 1992; Bardovi-Harlig & Bergstrom, 1996; Salaberry, 1999;
Kihlstedt, 2002; Howard, 2004 among others). The spread of imperfective over verb
classes is not as extensive as the spread of perfective/past. Even when the imperfective
reaches its last stage, achievements, the rates of appropriate use (accuracy) are higher
with the prototypical uses (atelic verbs) than with the non-prototypical uses (telic verbs).
Tokens of states in interlanguage are dominated by be and have or their equivalents. This
phenomenon is not seen in the acquisition of perfective past morphology.
The overgeneralization of progressive to states is observed in learner
interlanguages, but the rate is very low for learners with even limited instruction
(generally around 3-4% for personal narratives, and 7-9% for cloze passages) (Bardovi-
Major Studies on the Aspect Hypothesis
Some major studies on languages other than French and Chinese and their findings
are summarized below (studies on Chinese and French will be introduced in detail in the
Bardovi-Harlig (1998) used oral and written narratives from L2 learners of English
to examine the Aspect Hypothesis. In her study, the oral data supported the Aspect
Hypothesis better than the written data. The oral data showed a clear progression of past
tense from achievements to accomplishments to activities, whereas in the written data the
achievements and accomplishments patterned the same way. The written cloze passages
used by Bardovi-Harlig and Reynolds (1995) did not show a difference between the two
telic verb categories (accomplishments and achievements) either. Instead, the marking of
simple past was found to associate much more closely with telic verbs (achievements and
accomplishments) than with atelic verbs (activities and states).
Robison's results (1993 cited in Bardovi-Harlig, 2000) convincingly supported
the claims of the Aspect Hypothesis. Spanish-speaking learners of English performed oral
interviews, and the findings showed that event predicates (achievements) had the highest
use of simple past of all the aspectual categories. The rates of use of simple past tense
increased for all lexical categories when proficiency increased. Robison (1995) found that
the association of progressive marking with activities strengthened with proficiency level,
even as the verbal morphology developed with proficiency level. Also, the use of present
tense with states was observed in the past imperfective context.
The German learners of English in Rohde's (1996) study showed that for both
regular and irregular verbs, most of the verbs that are inflected belong to the category of
achievements. Bayley (1994) had similar results to Rohde, and showed that although
phonetic constraints determined the likelihood that a verb would be inflected for past
tense, the tendency for telic verbs (accomplishments and achievements) to carry past cuts
across phonetic categories. In other words, telic verbs are more likely to be marked with
past morphology, whether they are regular or irregular.
Salaberry claimed in his 1999 study that the effect of lexical aspect may not be as
prevalent during the beginning stages as it is during more advanced stages of L2 Spanish
acquisition. Yet, a reanalysis of the data in Salaberry (1999) by Bardovi-Harlig (2000)
showed that the data do follow the developmental predictions made by the hypothesis.
Based on the data collected from intermediate and advanced learners of Spanish using a
written cloze test, Salaberry (2002) found that the effect of the lexical aspect of the verb
does not appear to be stronger among advanced learners than among intermediate
learners. However, he claimed that it is possible that an even stronger effect of a single
marker of past tense (i.e., a default tense marker) could be detected in earlier stages of
Housen and Rohde provide counterexamples to the Aspect Hypothesis. Housen
(2002) reported a longitudinal study of a child with L1 Dutch learning English. While the
data on the acquisition of the progressive aspect supported the predictions made by the
Aspect Hypothesis, with the acquisition of simple past marking, the picture was not clear.
States (mostly irregular verbs) appeared early in the stages of development and much
more often than expected. In addition, an analysis of the types instead of the tokens
revealed a lack of the early association between achievements and simple past marking
proposed by the hypothesis. Housen argued that the Aspect Hypothesis might work only
for regular verbs, not for irregular ones.
Rohde (2002, p. 209) found from his earlier (1996) data of German children
learning English that states had a very high past-tense marking frequency rate: 80%-
100% among four children in obligatory past contexts.
Many fewer studies have been done on non-Indo-European languages (Bardovi-
Harlig, 2000). Shirai (1995 cited in Bardovi-Harlig, 2000) investigated the interlanguage
of three Chinese learners of Japanese at a Japanese language program. The results
provided evidence for an association of the past-tense marker -ta in Japanese and
achievements. These learners also showed dominant use of progressive -te i- with
activities. Chinese learners of Japanese in Shirai and Kurono (1998 cited in Bardovi-
Harlig, 2000) performed a grammaticality judgment task and were more accurate in
recognizing the correctness of -te i- with activities than with achievements.
Though the Discourse Hypothesis is not the hypothesis tested by the present study,
it is important that readers have some knowledge of it to better understand some of the
studies reviewed in the following two sections.
Cross-linguistic investigations have suggested that the distinction between
background and foreground is a universal of narrative discourse. In the early 1980s,
scholars such as Hopper (1982), Flashner (1989) and Kumpf (1984) suggested that a
relationship exists between the use of verbal morphology in interlanguage and the
grounding of the narrative. Their analysis and findings were later summarized as the
Interlanguage Discourse Hypothesis, which predicts that learners use verbal morphology
to distinguish foreground from background in narratives (Bardovi-Harlig, 1994). As to
which morphological markers are associated with which grounding, different studies
provide different results (Kumpf, 1984; Veronique, 1987; Bardovi-Harlig, 1995, 1998).
The most typical connections are between perfective markers and foreground as well as
between imperfective markers and background. Although the Discourse Hypothesis and
the Aspect Hypothesis seem to be competing (they both deal with the distribution of
verbal morphology) and distinct (one concerning narrative structure, one lexical aspect),
they both rest on shared features of temporal semantics. There are overlaps between
features that determine grounding (sequentiality, punctuality and completeness) and those
that determine lexical aspects (telicity, dynamicity and punctuality), which explain why
both hypotheses can often be supported by the same data in acquisition studies.
Andersen and Shirai (1994) also recognize the discourse motivations for the
distribution of tense-aspect morphology observed in L1 and L2 acquisition as well as
among native speakers. They argue that both learners and native speakers have the same
communicative need to "distinguish reference to the main point/goal of talk from
supporting information, within the tradition of research on grounding and the functions of
tense-aspect marking in narratives" (p. 152). The role of discourse is offered by Andersen
and Shirai more as an explanation to the Aspect Hypothesis than as another (competing)
hypothesis describing the distribution of tense-aspect morphology.
Studies on the Acquisition of French as a Second Language
As in other languages, studies on the acquisition of temporality in French focus on
the acquisition of pastness, in other words, the focus is on how learners acquire the notion
of past using different linguistic as well as extralinguistic means. Imparfait and Passe
Compose are the two main past tenses used in spoken French and they remain the center
of all the studies on the acquisition of temporality in French. Depending on the level of
the subjects, the focus shifts from verbal morphology, to discourse and pragmatics.
Studies have been done in various learning environments including untutored immigrant
workers, immersion students and beginning, intermediate and advanced foreign language
Trevise (1987) analyzed three short narratives from two Spanish-speaking
informants learning French using an analytical frame he proposed. This analytical frame
includes two main parts: the reference point and the structure of the narrative. Basically,
the analytical framework calls for a detailed investigation of all the possible ways of
expressing temporality, from verb morphology to world knowledge. The results indicated
that temporal linguistic markers and the notional value of time do not necessarily match
in learners' narratives. One notion can be represented by several grammatical or lexical
means; at the same time, one grammatical marker can refer to several different notions.
For example, one informant, Ml, did not always assign the same value to the only
temporal conjunction he used; quand (when) was used to mean when, before, after, and
at the same time. There were no meaningful opposition in verb morphology used by the
Trevise pointed out that the terms "foreground" and "background" do not represent
linguistic concepts and he used two other terms within "structure of narrative"
(mentioned above) instead: story line, on the one hand, and comments, explanations, and
descriptions, on the other.
Veronique (1987) designed a cross-sectional study to examine conversational
excerpts from Arabic- and Berber-speaking workers living in Marseilles, France. The
findings showed that the foreground-background dichotomy is too simple for
interlanguage analysis. In fact, his results indicated that verb morphology in
interlanguage was not determined by grounding, contrary to how it works in the target
language. Veronique suggested taking into account local constraints such as adverbials,
chronology and spatial devices also.
In addition, Veronique found that learners in his study used similar devices to those
of native speakers, namely:
* Reliance on the discursive principle: first happened, first mentioned
* Reliance on shared knowledge of the world and asyndetic relations between clauses
* Use of calendrical expressions and spatial reference
* Use of indexical and anaphoric adverbials
* Use of an elementary V stem (aux.) V + [e] verb morphology contrast (p. 267)
The European Science Foundation (ESF) project examines the acquisition of
temporality by foreign immigrant workers in industrialized Western European countries.
The target languages (TL) are English, German, Dutch, French and Swedish, while the
source languages (SL) are Punjabi, Italian, Turkish, Arabic, Spanish and Finnish. The
research involved in the project is cross-linguistic and longitudinal.
In the acquisition of French, four informants from two different SLs (Spanish and
Arabic) were interviewed. In each SL, a pair of two learners, one a slow learner and the
other a relatively fast learner, was studied in order to obtain insights on developmental
sequences. Each learner's production was analyzed in great detail, as is the case in
individual case studies. To avoid the inflectionall paradigm bias" (Klein, 1995), the
factors taken into consideration include not only tense and aspect but also temporal
adverbials, discourse organization and the principle of natural order1. The theoretical
model adopted in the project is a discourse-based functional model that the researchers of
the project call the "conceptual approach"(Dietrich & Perdue, 1995), which means that
the starting point is the concept "time". What is under investigation are all the sub-
components available to the learners to express the concept of time and the development
of these components over time.
Since the learners were all untutored (even though some of them had had
instruction in French for a very short period of time), their verbal morphology was very
underdeveloped. There were very few signs of systematic morphological distinctions and
these were idiosyncratically marked in relation to the TL when they emerged. For
1 The Principle of Natural Order (PNO, Klein, 1986) claims that narratives follow chronological order and
that deviation from chronological order must be indicated.
example, one of the learners of French, Berta, from a Spanish-speaking background,
seemed to make a mood distinction (realis vs. irrealis) with the verb pattern: V -[e] vs. V-
0. What the learners relied on was the discourse structure, in combination with a rich
repertoire of temporal adverbials. The principle of natural order was systematically used
from the first stage. The functions of adverbials were stretched beyond their target-like
uses to compensate for the lack of knowledge in verbal morphology. For example, deictic
adverbials were used as anaphoric ones. Another characteristic of the interlanguage was
the reliance on (framed and unframed) quoted speech as it provided available deictic
adverbials that learners could use with a shifted temporal value.
The objective of the ESF research was not to test the Aspect Hypothesis;
furthermore, it was not possible to test the Aspect Hypothesis, because the verbal
morphology was at its very early stages in these learners who participated in ESF.
However, one of the findings was in line with the results reported in many other studies
on the acquisition of French: although French has a grammaticalized aspectual distinction
in the past, there is no evidence in the data that even advanced learners acquire it.
The results also indicated that in learners' narratives, events in the foreground
tended to have an inherent boundary, and background situations tended to be durative and
to happen at the same time as some foreground events. This finding seems to agree with
the Discourse Hypothesis, in that the learners seem to distinguish the foreground from the
background, though not by means of verbal morphology. The devices associated with the
two groundings differed from one learner to the other. Similarly, Veronique (1987)
claimed that his study on African workers' French shows that the use of IL verbal
morphology was not determined by backgrounding or foregrounding as is the case in the
It is true that the functional approach is appropriate for the studies on untutored
learners because their underdeveloped verbal morphology prevents us from making any
quantitative claims, but this approach has several drawbacks. First, because it takes
everything into account, the analysis seems to be extremely long yet without focus. Some
of the description is so detailed that it requires adequate knowledge of the TL to
understand it. Second, there seems to be a lack of systematicity in the analysis. It is
sometimes hard to follow, as patterns are not clearly stated by the researchers. Third, the
detailed analysis requires a very limited number of subjects, which in turn limits the
generalizability of the results.
In the effort to enhance the grammatical competence of the immersion students
with respect to their use of the Imp and PC, Harley (1989) did an experimental study to
determine the effect of focused instruction on the acquisition of the two tenses. Activities
were designed to provide focused L2 input that promoted perception and comprehension
of functional contrasts between the two tenses and to provide more opportunities for
students to express these functions in the realization of interesting, motivating tasks.
Students were given pre-tests, immediate post-tests and delayed post-tests using
compositions, cloze tests and oral interviews. Only immediate benefits were found
among the students in the experimental group. In the long run, the experimental students
did not do significantly better than the students in the control group. Nevertheless, the
present author does not think that it means the treatment was not successful; the fact that
near-native speakers have trouble in grammaticality judgments regarding PC and Imp
simply shows that we cannot expect to teach this notorious feature of French in 8 weeks.
Harley (1992) conducted a cross-sectional study to seek patterns of second
language development in the acquisition of French by English-speaking immersion
students in Canada. Thirty-six early immersion students (from Kindergarten), 12 late
immersion students and 24 native speakers were interviewed. The grade levels of the
students were grades 1, 4 and 10.
Among the various features of the French verb system under investigation were
tense and aspect. The findings suggested a relationship between the tense chosen and the
lexical aspect of the verb. PC was found almost exclusively with dynamic verbs whereas
Imp was lexically restricted to a small set of inherently durative verbs such as aimer,
avoir, &tre, pouvoir, savoir and vouloir.
As the subjects in the ESF project, the grade 1 students were found to rely on the
wider discourse context or accompanying adverbs, rather than using verbal morphology,
to convey distinctions in time. This suggests a universal early stage of acquisition where
lexical and discourse means play a more important role than verb morphology.
It seems that Imp poses more difficulty for the students than PC. Grade 10 students
(10 years of immersion beginning in kindergarten) produced regular use of Imp to refer to
past action in progress, but very little use of this form in the context of habitual past,
where Imp is obligatory. A study of adult advanced learners of French (Kihlstedt, 2002)
confirms the relative difficulty of the acquisition of this function of Imp. Several reasons
are proposed by Harley to account for the slow development of Imp, among which are the
multiple homophony of Imp with the infinitive and second person plural/imperative, the
optionality of its use in many contexts and its application, unlike the English progressive,
to both states and actions in the past. But note that the students use Imp with durative
verbs as aimer, avoir, &tre, pouvoir, savoir and vouloir, which in English do not normally
occur in the progressive. The author herself states in the conclusion that those lexically
bound instances of Imp might be unanalyzed chunks.
Beginning and intermediate foreign language learners
Bardovi-Harlig and Bergstrom (1996) tested the four effects of the Primacy of
Aspect Hypothesis (the Aspect Hypothesis) using acquisition studies of English and
French as second/foreign language. Twenty-three ESL and 23 FFL learners were asked to
recall an 8-minute excerpt from the silent movie Modern Times. The results indicated that
the acquisitional patterns in both English and French show effects of lexical aspect. In
other words, with regards to French, Imp appeared later than PC; Imp marking began
with states, extending next to other verb classes; the spread of PC had an early strong
association with achievements, followed by accomplishments before spreading to
activities and sates. But as in Harley (1992), Imp was also found to be lexically restricted
to several durative verbs.
Bardovi-Harlig suggested that language teachers provide contextualized examples
of state predicates with Imp in French since Imp seemed to be undergeneralized, with
avoir and &tre accounting for 81.3% of all the instances of states used in the Imp. This
seems to be a cross-level problem since even the use of Imp by the advanced learners in
Kihlstaedt's (2000) study and of the immersion learners in Harley's (1992) study was
restricted to a limited number of durative verbs. One example from Kihlstedt illustrates
the point very well: one of the subjects, Marie, identified as a late advanced learner,
hesitated and asked for confirmation when extending Imp to the less frequent state verb,
Kaplan (1987) used error analysis in her study of the developmental patterns of past
tense acquisition among first and second year university learners of French. The purpose
of the study was to test the Input Hypothesis to see what effects, if any, instruction had on
the observed generalities. Results showed that the use of PC has a greater overall
accuracy rate than Imp. In addition, the verb types2 supplied for PC was four times
greater than those supplied for Imp. Imp was clearly underrepresented in these learners'
output. Imp was often presented by a present tense form, while relatively few present
tense forms occurred in the PC slots.
In trying to explain the greater accuracy rate of PC, Kaplan rejected the effect of
instruction, because Imp was introduced earlier than PC for those learners. This does not
stand as a valid explanation because introducing first is far from being a strong condition
for a greater accuracy rate. Besides, the role of instruction implies much more than the
order of introduction. The author also proposed the greater frequency of occurrence in the
classroom of PC and the subtle aspectual notion of the Imp to explain the results.
Kaplan also claimed that the precedence given to (grammatical not lexical)
aspectual organization over time-marking may be a universal language acquisition
feature. Support of this claim comes from the observation that present tense was used by
the learners in place of Imp but not PC, indicating that learners did see the distinction
2 Verb types are different from verb tokens which include every occurrence of a verb in any form. Verb
types are instances of a verb with different inflectional endings. For example, two instances of avoir in first
person singular count for one type but two tokens.
between Imp and PC. As the distinction was not temporal, it had to be aspectual. There is
no consensus in the literature whether the learners mark aspect or tense first in their
acquisition of temporality.
Since the basis on which the Aspect Hypothesis was proposed is L1 acquisition,
Salaberry (1998) posed the following question: should we assume that adult classroom
learners will develop their L2 grammar in the same way as an adolescent in the natural
environment? To answer the question, Salaberry conducted a study to analyze the
development of aspect markers (PC and Imp) in 39 English-speaking second-semester
learners of French and in their control group, 30 native speakers. The data elicitation
tasks involved a written narration of a short film and a cloze test. Results from both tasks
showed that the selection of past tense marking by native speakers and learners coincided
with respect to the prototypical relations between grammatical aspect and inherent
semantic values. The data from the cloze test revealed that learners differed from native
speakers in their use of non-prototypical grammatical aspect in L2 French. The statistical
tests of the narrative task in the study revealed significant differences in the use of verbal
morphology between telic verbs (achievements) and atelic verbs (statives and activities)
for both groups.
Salaberry viewed the extended use of the PC in both tasks among the learners as a
piece of evidence to consider PC a default marker of past tense. The free written task
(written narratives) revealed more clearly non-native speakers' over-reliance on the use
of this potential default marker of the past tense.
Because the data from the cloze test showed that classroom learners were not
successful in the selection of the non-prototypical forms associated with states (PC),
Salaberry concluded that the spread of PC and Imp from prototypical to non-prototypical
forms might differ between academic and natural learners. His reasoning was that
classroom learners did not have access to a highly contextualized linguistic environment.
While it is plausible to account for the difference between classroom learners and natural
learners according to the different linguistic environments they are in, it is not clear how
Salaberry reached the conclusion that the spread of tense-aspect morphology differed
between the two. He did not make it clear whether the difference existed in the
distribution of PC and Imp over lexical aspects or in the accuracy of the use of PC and
Imp or both.
It was likewise not clear how his study showed a different trend than the
developmental sequences proposed in the Aspect Hypothesis, although it was not a
longitudinal or cross-sectional study. He seemed to think that a similar profile of use of
verbal morphology according to aspectual class between native and non-native speakers
would suffice to disprove the developmental sequences. The problem with his claim is
that he only had learners from one level; the fact that their use of verbal morphology over
aspectual classes pattern similarly with that of native speakers does not mean that they
did not go through the developmental stages proposed by the Aspect Hypothesis.
Advanced foreign language learners:
It is true that research on advanced learners is sparse compared to that on beginning
and intermediate learners. Kihlstedt (2002) tried to fill the gap by investigating the
acquisition of tense and aspect by advanced learners of French. His study investigated the
following question: Are advanced learners of French still more sensitive to the
distributional bias of the input so that the prototypical associations of Imp and PC are
more strongly reflected by the learners than by the native speakers?
Generally speaking, the learners in this study were not more constrained by
inherent lexical aspect than native speakers. The spread of different verb types in PC and
Imp was compatible with that of the native speakers; Imp cluttered predominantly in
states, both in the learner group (88%) and the native group (85%). But a "correct" form
used "appropriately" in interlanguage can be both correct and appropriate for the wrong
reasons. As Kihlstedt pointed out in his discussion: form precedes function. When past
inflections begin spreading to the non-prototypical combinations, such as telic verbs in
Imp and states in the PC, they do not necessarily express the same functions as in native
Most of the mistakes that the learners made in Kihlstedt's study were associated
with the use of Imp. When learners used Imp on their own initiative, the tendency was to
explicitly mark duration. In fact, Imp is normally incompatible with time adverbials that
express a quantified, limited duration, except when the situation can be seen as habitually
repeated within a global temporal frame. The late acquisition of this particular function of
Imp: Habitual Imperfect, has also been observed in English-speaking immersion students
(Harley, 1992). To explain the problematic feature of the Habitual Imperfect, Kihlstedt
proposed that in this function, Event Time and Reference Time are separated. The present
author suspects an additional explanation, which is that for English speakers, the Habitual
Imperfect is expressed by an analytic form "used to", so it is hard for the students to
accept a synthetic form. The same explanation works for the lack of use of Imp to express
the progressive, which is expressed in English by the analytic form be+ -Ving. Kihlstedt
further claimed that the principle semantic division of inflection was +/- dynamicity
rather than +/- telicity.
Kihlstedt proposed that the default relation of reference to past time is before now,
and the more an event is specified, aspectually or temporally, the higher the acquisitional
stage. This explains in part the acquisitional difficulty of Imp.
Howard (2004) tested both the Aspect Hypothesis and the Discourse Hypothesis
with three groups of advanced learners of French. The results of the study showed that
the use of each morphological form varied not only between discourse grounds, but also
within each discourse ground as a function of the lexical verb class. Howard concluded
that it was the joined force of both lexical aspect and grounding that determined tense-
aspect marking by learners. In fact, Howard claimed that many factors besides lexical
aspect and grounding (e.g. phonetic salience, verb irregularity) interacted to create the
interlanguage, and that future research should investigate multiple causes of the
variations in tense-aspect marking.
Studies on the Acquisition of Chinese
Research on the acquisition of temporality in Chinese is focused on the acquisition
of aspect (grammatical aspect). This focus can perhaps be explained by the fact that
aspect is the only category in Chinese temporality that is grammaticalized as in Indo-
European languages. What's more, in the aspectual category, the particle LE (verb final -
le more than sentence final le) receives much more attention than other particles because
it is the most frequently used aspect marker in the language and it has been found to pose
the most problems to learners.
There are far fewer acquisition studies on non-Indo-European languages than on
Indo-European languages; studies on Chinese are especially sparse. There are very
limited numbers of studies available on the acquisition of temporality in L1 and L2
L1 Acquisition of Chinese Temporality
Erbaugh was the first to study the acquisition of temporality in Mandarin Chinese.
The subjects of her study (1985), two two-year-old Mandarin-speaking children, virtually
never marked time or aspect; the only exception was their broad use of the perfective
verb suffix le.
The features that triggered the perfective were found to follow the ranking below
from the most important to the least important: pastness of event, clear end point, re-
enactment potential, transitivity and agentivity. The first two factors are self-explanatory.
Re-enactment potential "expresses the degree of sensory support which the child has for
rehearsing the event she wants to describe"(p. 59). Agentivity means that the situations
described involve the children themselves as agents. Transivity indicates the "potency
and affectedness of the agent and patient" (p. 60).
Inherent lexical aspect was found to be closely associated with the marking of -le
under the category clear end point. Punctual aspect was very strongly correlated with -le
use; more durative verbs could also be used to express a clear-cut end point and Chinese
Resultative Complement Expressions played a critical role in marking end points on
durative actions. In addition, as mentioned in Chapter Two, Resultative Complement
Expressions (Resultative Compounds in Chapter Two) are a form of achievements. Thus,
the results in Erbaugh seem to support the Aspect Hypothesis in that the perfective -le is
more strongly associated with telic verbs (achievements and accomplishments) than with
atelic verbs. Both marked and unmarked past references used a full semantic range of
predicates, including actions, processes and states, thus supporting Weist's (1984)
assertion that past marking appears cross-linguistically for a full range of predicate types.
Erbaugh (1985) stated that the five factors were universal and that they applied to
both children and adults. More importantly, he concluded that personal involvement was
a far more powerful trigger for aspect marking than syntactic or semantic factors for
Li & Bowerman (1998) designed a very rigorous experimental study on the
acquisition of aspect particles: le, -guo, -zhe and zai by 135 Mandarin-speaking children
aging from four to six years old. They proposed that there were natural associations
between lexical and grammatical aspect in Chinese: between the perfective marker -le
and accomplishments/achievements, between the progressive marker zai and activities,
and between the durative marker -zhe and states. The purpose of their study was to see if
children were sensitive to those natural associations.
The results of the experiments suggested that the Chinese children were indeed
sensitive to these associations and more analysis showed that there was a consistent
association of imperfective markers with atelic verbs and of the perfective marker with
telic verbs. Statives and semelfactives3 patterned in general like activities. The authors
concluded from those results that children were indeed sensitive to the distinction
between process and result as proposed in the Basic Child Grammar Hypothesis (Slobin,
1985 cited in Li & Bowerman, 1998). There was, however, no evidence for the
Bioprogram Hypothesis (Bikerton, 1981 & 1984 cited in Li & Bowerman, 1998), which
states that distinctions between state and process and between punctual and nonpunctual
3 Smith (1991) added semelfactives as an expansion of Vendler's four-way verb classification.
Semelfactives are verbs such as 'cough', 'tap' and 'knock'. They are similar to achievements as being
punctual. When combined with progressive markers, they are interpreted as specifying a repeated situation.
are preprogrammed into learners. Li and Bowerman pointed out that future research
should not focus on the preprogrammed form-meaning mapping in tense-aspect
acquisition, but, rather, should seek alternative explanations in models that emphasize
children's ability to detect patterns through the formation of prototypes.
Huang (2000) investigated how two Mandarin-speaking children referred to the
past, taking into account not only aspectual particles, but also semantic and discourse-
The two children's references to the past were divided into two categories: with or
without overt markers. The analysis of the tape recording showed that when children
marked their past references, they resorted mostly to aspect markers and most frequently
used the perfective marker -le. Experiential marker guo was used for distant past
reference. Both guo and -le were used by the children to indicate deictic temporal
relations where the reference was speech time.
In line with the findings in Erbaugh (1985), Huang found that children's access to
temporal adverbs appeared to be rather limited, as opposed to their mothers, who used
mainly temporal adverbs to establish past time.
In Huang's study, the children used the perfective marker -le predominantly with
resultative verbs (Resultative Compounds), therefore, it appears that their use of the
perfective marker was associated with the inherent lexical aspect of verbs. It was also
stated in the study that early perfective marking was strongly associated with the sense of
pastness. Huang resorted to the prototype account proposed in Andersen and Shirai
(1994) to explain the influence of lexical aspect on the acquisition of grammatical aspect.
The reason why resultative verbs and pastness were associated with perfective marking is
that these two features are among the prototypical features of the perfective aspect:
[+unitary], [+result state], [+punctual] and [+past]. Huang also mentioned that the
situations described often involved the children themselves as agent, which Erbaugh
termed "agentivity". For instances where past references were unmarked, context, shared
knowledge and world knowledge helped establish the time reference between children
and their mothers.
L2 Acquisition of Chinese Temporality
According to Duff & Li (2002), the difficulty of acquiring perfective -le 4for L2
learners lies in the multifunctionality of -le, its interaction with the inherent lexical aspect
of predicates, and the role played by the speaker's perception or viewpoint of the relative
boundedness of an event. The multifunctionality of -le is illustrated in this study by a
decontextualized sentence, along with its five different possible interpretations. For
learners from a tensed language, it is very hard to image an inflected verb in a simple
sentence having five interpretations. The sentence given in the article, along with its five
interpretations, are reproduced here (p. 421):
Ta lai-LE (ambiguous)
i. He came
ii. He has come
iii. He is coming
iv. There he comes
v. After he comes
4 The perfective -le in Duff and Li's study excluded typical sentence-final le such as Adjective+le, but did
not differentiate them in overlapping instances where verb-final -le occurred at the end of the sentence.
The functions listed by Duff and Li as well as those listed by Wen (1998) do not
seem to have relationships among them. In other words, the different functions of -le
seem scattered in the language, so it is very difficult for L2 learners to grasp the gist of
-le. Moreover, as pointed out by Duff and Li, aspect marking is very much oriented
toward discourse/pragmatics and, in many cases, is syntactically optional according to the
Nine native speakers (NSs) and nine non-native speakers (NNSs) of Chinese
completed two oral activities and one written editing task. Based on the results of these
activities, it was observed that there are several factors responsible for learners'
production (or omission) of the perfective verb-final -le in Mandarin: L1 transfer from
English, cognitive factors related to the functional load of-le, learner's exposure to
Chinese, learner's proficiency, inherent lexical aspect, discourse features of the tasks, and
the effect of instruction and textbook explanation.
The results showed that NNSs, particularly those with low proficiency levels,
tended to undersupply -le in their oral narratives, and tended to oversupply it with certain
state and non-perfective activities. There were also trends including subjects' non-
suppliance of -le at early points of acquisition, then oversuppliance (attributed to
overgeneralization and transfer of L1 tense-marking-sensitivity). The initial instruction
and awareness that Mandarin does not have tense, combined with early interlanguage
simplification, initially allowed NNSs to suppress all grammatical marking for past
and/or perfective events. By the second semester of university study, and increased
exposure to -le, students begin to consciously produce, and even overproduce -le, with
quantified or other specified objects.
Most of the NNSs' correct suppliance of -le in the written editing task co-occurred
with accomplishments/achievements with quantified or specified objects; incorrect
suppliance occurred with states, "say" and "think" verbs, or verbs in nonperfective
Though not directly, the authors also discussed the role of grounding when talking
about the influence of the task types. The personal narrative of vacation travel did not
elicit as many -les from NSs as the oral video-story-telling, because the travel narratives
involved more background description.
One interesting finding in the study is that one of the reasons that the NNSs did not
use as many -les as NSs is that they did not have enough vocabulary. The authors posited
that NNSs had a smaller repertoire of inherently perfective verbs (V, V+ Quantified
Objects or Resultative Compounds), so they attached -le to more generic and less
prototypically perfective verbs.
In a study conducted by Wen (1995), fourteen English-speaking students of two
different proficiency levels took part in three oral interviews. Wen divided LEs into two
categories: verb suffix -le and sentence-final le. He claimed that the two are functionally
different and are learned differently in L2 acquisition. Verb suffix -le has a perfective
aspect function, whereas sentence-final le signals the end of the sentence and change of
state. Each utterance ofLE was coded as verb-final or sentence-final and the accuracy of
each category was calculated. While most of the studies acknowledge the differences of
the two LEs, they focus on the perfective marker or, as in Duff and Li, on the overlapping
instances, such as when verb-final -le occurs at the end of the sentence.
In other studies that differentiate the two LEs, Erbaugh (1985) found that 80% to
90% of the cases of LEs by native Chinese-speaking children are perfective markers,
leaving only 10% to 20% sentence-final. The findings of Wen's study agree with what
was observed by Erbaugh in L1 acquisition: English-speaking learners acquire the
perfective aspect -le earlier than sentence-final le. The evidence makes Wen believe that
sentence-final le contains certain linguistic and non-linguistic difficulties not associated
with the verb suffix -le. He posits that the constraints on verbs are more active than those
on sentences in the minds of both children and adult speakers. In addition, according to
Wen, the meaning of the verb suffix -le is more concrete and less varied than sentence
final le. Teng (1998) observed the opposite trends, showing that sentence-final le was
relatively uncomplicated and was acquired by L2 learners with certainty at a fairly early
stage of acquisition, while the verb-final -le was acquired with a persistent ratio of errors
perhaps throughout a number of years of the learning career.
Other results of Wen's study include: 1. the students at beginning and more
advanced levels do not differ significantly in their accuracy of the verb suffix -le, but
more advanced students are more often able to correctly use sentence-final le; 2. verbal
complements influence the accuracy of both Les. For example, students frequently
omitted sentence-final le when it was followed by a verb phrase with an object or a
resultative verb complement5; 3. the use of adverbs such as yijing (already) and tai (too)
decrease the accuracy rate in the use ofLE; 4. students usually used the verb suffix -le
5 Li & Duff (2002) however, suggest that the use of a resultative verb complement might help students
produce more les.
with verbs which had an inherent end point built into their meaning or when a durative
verb expressed a clear-cut end point.
Distributional Bias Hypothesis and Prototype Hypothesis
In his attempt to explain the observations found in L2 acquisition of tense-aspect
morphology, Andersen (1986 cited in Bardovi-Harlig, 2000) proposed the Distributional
Bias Hypothesis, which claims that the patterns found in learners' interlanguage also
exist in adult native speakers' speech. In other words, the biased distribution of tense-
aspect morphology across the four verb classes is a phenomenon also found among native
Before the Distributional Bias Hypothesis, there seemed to be a gap between
studies in theoretical linguistics on theories governing aspectual systems and studies in
SLA on acquisition processes, with each focusing on their own field. It is inevitable that
acquisitional studies should discuss how the aspectual system works in the language in
question, but few of them talk about the distribution of grammatical aspect markers over
lexical aspectual classes while at the same time investigating the relationship between the
two aspects in learners' interlanguage. It is very hard to follow the acquisitional studies
without a complete picture of the language, especially if the reader does not possess
adequate knowledge of the language under investigation. As Andersen (1993) points out,
the acquisitional studies assume an equal distribution of grammatical aspects across
lexical aspects in native speech. What's more, when some studies talk about incorrect
associations made by the learners (e.g. overextending progressive to states), they give the
false impression that certain associations between grammatical aspect and lexical aspect
are not allowed. The truth is, any grammatical aspect can be combined with any verb
type, while restrictions apply as to exactly which specific verb may be combined with
which specific grammatical aspect. For example, perfective aspect can occur with any
verb type, but it does not mean that it can occur with any single verb in that verb type in
In order to test the Distributional Bias Hypothesis, Andersen's students conducted
studies in several different languages including Japanese, English, and Italian (Andersen,
1993). Their findings are summarized below:
1. There is a greater proportion of the combined category of telic and punctual verbs
receiving past or perfective inflections than the other three categories;
2. There is a greater proportion of activities receiving the progressive inflection than
is the case for past, perfective and present forms;
3. A lower proportion of imperfective inflections occur on telic or punctual verbs than
is the case for perfective inflections.
4. Present tense inflection occurs most frequently with states.
The data elicitation in those studies is limited to interviews and there does not seem
to be an equally distributed number of different verb types (so the telic and punctual
verbs have to be combined together); those studies nevertheless prove that the
distributional bias exists in native speakers' speech.
In an attempt to account for the distributional bias found in learners' as well as
native speakers' data, Shirai and Andersen (1995) resort to the Prototype Hypothesis to
explain the underlying cognitive processes involved in the acquisition of tense-aspect
morphology. The Prototype Hypothesis was developed first in philosophy and in
cognitive psychology. It assumes a graded category membership. A category has its best
exemplars, which are the prototypes of that category, as well as other peripheral
members. For example, when we consider "furniture", tables, chairs, couches and beds
come more easily to mind than ottomans. When applied to the acquisition of languages,
the prototype theory predicts that in L1 and L2 acquisition, learners first acquire the
prototypical members of a grammatical category and gradually extend their knowledge to
the non-prototypical members.
In a prototype model, for the simple past tense in English, the core or more
prototypical function is to indicate past time reference, but we know that it can also
indicate counterfactual meaning, as in IfI were you, or be used as a pragmatic softener as
in Couldyou lend me some money? Even in the category that indicates past time
reference, there are hierarchies among the different verb classes. Because the features of
achievements (i.e. +punctual, +telic, and +result) overlap more with some of the features
of perfective aspect (or simple past tense), it is more likely for achievements to be
marked by the perfective aspect morpheme. In this sense, we say that there is a
prototypical association between achievements and the perfective aspect morphology.
The prototypicality decreases from achievements to accomplishments, then to activities,
with the least prototypical verb class being states. The above statement can be
summarized in the following hierarchy (Andersen & Shirai, 1996, p. 557) for English
simple past tense morphology from prototypical members to marginal members:
Deictic past (achievement- accomplishment- activity state habitual or
iterative past)- counterfactual or pragmatic softener
Following the same rationale, the English progressive can be represented by the
following hierarchy (Andersen & Shirai, 1996, p. 557):
Process (activity- accomplishment)- iterative- habitual or future stative
Parts of these two hierarchies are actually the claims of the Aspect Hypothesis for
the acquisition of past tense or perfective aspect morphology and progressive aspect. The
claim regarding the distribution of imperfective grammatical aspect across verb classes
can be represented by the following hierarchy:
State activity accomplishment achievement
One question arises naturally: how do we determine the prototypical associations in
language acquisition? They could be inferred based on the core meanings (language
universals) of a grammatical category and the specific meanings of the category in the
language in question. But as Andersen & Shirai (1996) point out, there is no established
and reliable measure to determine the internal structure of a prototype category.
Researchers' intuition and psycholingustic experiments are the only ways to determine
the associations. However, to use acquisitional experiments to determine patterns of
acquisition seems circular, and this is the weakness of the Prototype Hypothesis in
The objective of the present paper is to test the Prototype Hypothesis using French
and Chinese L2 data to see if it can explain the observations found in the data from both
languages. What is under investigation is the relationship between lexical aspect and
grammatical aspect in the two languages. While testing the Aspect Hypothesis requires a
longitudinal or at least cross-sectional study, testing the Prototype Hypothesis can be
accomplished with learners all at one level (determined by their course level), which is
the design of the present study. The Prototype Hypothesis predicts that there will be a
biased distribution of tense-aspect marking over four aspectual classes in the
interlanguage. As far as the acquisition of French and Chinese is concerned, the relevant
1. Perfective marking (PC in French, -le and -guo in Chinese) is more closely related
to achievements and accomplishments than to activities and sates.
2. Imperfective marking (Imp in French) is more closely related to states and activities
than to accomplishments and achievements.
3. Progressive marking (zai in Chinese) is more closely related to activities than to
accomplishments and achievements.
More specifically, the present study will consider the following research questions:
1. Is the use of imparfait and passe compose by intermediate-level learners of
French as an L2 influenced by the inherent semantic aspects of verbal predicates?
* Does the frequency of the use of imparfait and passe compose vary depending on
the verb class considered?
* Does the level of accurate use of imparfait and passe compose vary depending on
the verb class considered?
2. Is the use of the grammatical aspect markers (-le, -guo and zai) by intermediate-
level learners of Chinese as an L2 influenced by the inherent semantic aspects of verbal
* Does the frequency of the use of the markers vary depending on the verb class
* Does the level of accurate use of the markers vary depending on the verb class
3. How does the acquisition of grammatical aspect in the two languages differ in
terms of the influence of the lexical aspect?
The objective of the present study is to examine the relationship between inherent
lexical aspect and tense-aspect morphology in learners' production. More precisely, it is
to test the Prototype Hypothesis in order to see the distribution of the tense-aspect
morphology across four aspectual classes: states, activities, accomplishments and
achievements. The present study also looks at the comprehension of time reference in
general by the learners. The languages under investigation are French and Chinese, and
past time reference will be the focus for the following reasons:
1. In most languages, only the past time reference is morphologically marked, because
in order to indicate that the time reference is different from the 'here and now',
some sort of morphological marker is required. Future time reference is not a good
candidate, because it often involves the distinction between realis and irrealis,
which goes beyond the focus of the present study.
2. The perfective/imperfective opposition exists in French only in the past time
The participants in this study were students enrolled in 3rd (French) and 4th
(Chinese) semester foreign language classes at University of Florida. The students
learning French were from two sections taught by two different instructors, while the
students learning Chinese were from two sections taught by the same instructor. All
students participated in the study on a volunteer basis. On the French side, 26 students
completed all the activities, but one was excluded from the study because he came to the
United States after the age of 18 from a Spanish-speaking country and thus English was
not his first language; on the Chinese side, 20 students completed all the tasks. All the
students were native speakers of English; some of them were bilingual while some others
knew more than one foreign language. There were no students learning French who were
native speakers of French, but there were 7 students learning Chinese who encountered
spoken Chinese (either Mandarin or Cantonese) at home before English and have parents
who are native speakers of Chinese. It should be pointed out that none of them can be
considered 'bilinguals' because they cannot even hold a simple conversation in Chinese
according to their teacher (they are called 'heritage students' in the following table) and
the language they use at home is English. Only the data from the students who did all the
activities were included in the data analysis.
The backgrounds of the students are summarized in table 4.1:
Table 4.1 Information on the participants
Number of participants 25 20
Number of bilinguals 8 (Spanish or Creole) 9 (Spanish or Chinese)
Males 5 11
Females 20 9
Number of students who 5 2
have studied in target-
Number of heritage students 0 7
French and Chinese Programs
The elementary French program consists of one year, two semesters. The second
year students are considered to be at the intermediate level. The students are placed in the
3rd semester course based on the successful completion of the first year courses or by
their SATII score. They meet three times a week for a grammar class, and in addition,
twice a week for a conversation class to practice speaking. The data were collected from
the grammar class. A communicative teaching approach is used in the classroom, where
the target language is used more than 90% of the time.
Like the French program, the elementary Chinese program consists of one year,
two semesters. The participants in the present study are at the 4th semester of their
Chinese study and are also considered to be at an intermediate level. The reason for
which the participants in the Chinese program were one semester ahead of their French
counterparts is that the contrast between Chinese and English is far greater than that
between French and English, making Chinese much more difficult for English speakers
to acquire. Chinese writing system, which is not alphabetic and thus has nothing to do
with the way it is pronounced adds on to its difficulty for English speakers. In fact, in the
French program, students start writing compositions from the first exam in their first
semester, while in the Chinese program, the students never really have to write a
composition in their exams until the second year. They do have a composition part in
their first year exams, but the topic is given ahead of time and students practically rewrite
what they have prepared outside of class. Since a major part of the data come from
compositions, it was predicted that fourth semester students would produce more
adequate data than third semester students, who just started to have free-writing
experience. In fact, in a pilot study, data was also collected from first year students
learning Chinese, but most of the students did not give enough material to allow for an
analysis of their compositions.
French and Chinese Textbooks
What is under investigation is the acquisition of the tense-aspect morphology in the
two languages. Let us first discuss how the content is presented in the textbooks the
students were using. As presented in the review of the literature, the
perfective/imperfective distinction exists only in past time reference in French. The
perfective past is represented by Passt Compost and the imperfective past Imparfait. In
the French textbook used by the students at UF, and in fact, in the majority of the
textbooks available on the market, PC is traditionally introduced first and Imp next.
Interactions is the textbook used by intermediate-level French learners at UF. The
textbook states that PC is used to express an action that was completed within a specified
or implied time frame in the past. One section in the explanation of the use of PC is
devoted to the non-prototypical combination of PC with state verbs such as avoir, &tre,
penser, vouloir, and pouvoir. It is explained that PC is used here because these state verbs
refer to an immediate reaction to an event or situation. Imp is said to have three
functions: to set a scene, to express habitual actions and to express a state or condition.
The textbook also contrasts PC with Imp. Imp is said to be used for description,
whereas PC is for the events happening in the foreground. The same verbs used in both
tenses in different sentences are presented and the nuances between the two uses are
explained with a translation of the sentences. Again, the same state verbs mentioned
above are chosen as examples to illustrate the nuances. The textbook repeatedly points
out to students that PC is not the only past tense used in French.
In Chinese, there is also a perfective/imperfective aspectual opposition. The
perfective markers are -le and -guo, while the imperfective markers are zai and -zhe. In
the textbook used by the participants, the four markers are not introduced sequentially,
nor as a single grammatical category. Zai, -le and -guo are introduced during the first
year and reviewed in the second year, while -zhe is introduced in the second year. Zai is
first presented as a proposition of location.
In the Chinese first-year materials in use, verb-final -le is first introduced as a
"dynamic particle" signifying realization or completion of an action or an event. It is
emphasized from the beginning that -le does not equal the past tense in English because
it can be combined with future situations. The grammar explanation and the exercises
both direct students' attention to a specific time adverbial and a quantified object that co-
occur with -le. It is noted that to negate -le, one should use "mei (you)" (not have)
instead of"bu" (no).
The discourse function of verb final -le is also mentioned in the textbook. It states
that when there is a series of actions or events, -le is usually used at the end of the series,
rather than after each of the verbs. Sentence-final le is introduced later than verb-final
-le. The book explains that sentence-final le indicates a change of status or the
realization of a new situation.
The structure where both verb-final -le and sentence-final le are used is also
discussed in the textbook. This structure is presented as "V+-le + Number + Measure
Word + Noun + le" and is contrasted with another structure which is the similar but
without the sentence-final le. The difference between the two is that in the first structure,
where sentence-final le is used, it is implied that the situation has been continuing and
will continue, while the same implicature cannot be deduced from the second structure,
where sentence-final le is not used.
At the beginning of the second year of classes at UF, the use of both verb-final le
and sentence-final le is reviewed. This time, the function of "relative past" is introduced,
in the sense that verb-final -le is used in the first situation of a series of two situations
that happen one after the other.
(Zheng) Zai is introduced in main clauses with "when" subordinate clauses and is
explained to express the "ongoing process of an action at a certain point of time".
Also called a "dynamic particle" in the textbook, -guo is introduced in the second
semester of the first year. It is said to denote a past experience, which did not continue to
the present but has an impact on the present.
The activities that the students completed include: a language background
information questionnaire (see appendix A) where they talked about the languages they
knew and their levels of each one; a comprehension task where the students were asked
to indicate the time reference of each verb predicate (see Appendix B) and a cloze test
(see appendix C) in French; a comprehension task (see Appendix D) and a written editing
task in Chinese (see appendix E) and three compositions on three different topics (see
On the language background information questionnaire, students were asked to
provide information concerning their native languagess, second languagess, the age at
which they started learning their second languages) and the length of study. Questions
on whether they had extensive experience studying or living in target language countries
were also asked. The objective of the language background information questionnaire
was to ensure that only native English speakers were included in the study. The Chinese
'heritage' students constitute a special group because their native language is either
Mandarin or Cantonese, but their primary language used both at school and at home is
English and, as mentioned above, none of them can hold a simple conversation in
Very few studies on the acquisition of tense-aspect morphology have used
comprehension tasks. The reason to include a written comprehension task was to gauge
the students' ability to comprehend the tense-aspect system before analyzing their
production. If problems are detected in production, it is helpful to compare production
with comprehension to see if the same kind of problems exists in comprehension.
Different sentences were included in the French and Chinese comprehension tasks
because the students did not have the same set of vocabulary. In the comprehension task,
students were asked to choose the time reference from the list for some underlined verbs:
a.- occurred in the past
b.- occurs in the present or is always true
c.- will occur in the future
A sample sentence, with four verbs, is given below from the French task:
Ce matin, nous sommes tous arrives a l'ecole bien contents, parce qu'on va
prendre une photo de la classes qui sera pour nous un souvenir que nous allons
cherir toute notre vie.
In the design of the comprehension tasks for both French and Chinese, the
instructors were consulted to make sure that the vocabulary was within the students'
range. For the French comprehension task, sentences were chosen from the book Le Petit
Nicolas. There are in total 20 verbs, which include instances used in the past, present and
future. Among the verbs used in the past, both uses of Imp and PC were present.
The present author wrote the sentences used in the Chinese comprehension task. As
in the French task, among the 26 verbs, there are instances in the past, present as well as
in the future. There are seven verbs that have aspect markers -le, -guo or zai. Because the
use of aspect markers is only one way of expressing time reference in Chinese, both
instances with and without these markers were included in the task to see if the
comprehension would be influenced by the markers.
The topics of the three compositions were chosen based on the content and, more
precisely, the verb morphology solicited in the compositions. Camps (2002) showed that
beginning language learner's use of verb morphology is influenced by the type of
narrative elicited. While the use of preterite does not seem to change with narrative type,
the frequency and accuracy of imperfect depends on narrative types: impersonal or
personal narratives. Personal narratives seem to generate a more balanced distribution of
perfective and imperfective. The three topics in the present study were chosen to elicit
personal narratives and are replicated from Camps (2002), which have been proven to
elicit desirable data:
Composition topic 1: Talk about what you and your family did during the winter
Composition topic 2. Talk about the things you used to do when you were 16 years
old. Give examples of specific events.
Composition topic 3. Talk about what you did last weekend, and compare that to
what you used to do on weekends when you were in high school.
The first composition sought perfective aspect markers, the second imperfective
markers and the third a combination of the two. The three topics made sure that there
would be contexts for all the aspectual markers tested and a relative balance between the
imperfective and the perfective. The topics for the French and Chinese students were the
same and were given in English. No dictionaries were allowed. Each composition took
approximately 25 minutes.
In French, there is a close relationship between habitual past situations and the use
of Imp: habitual situations in the past are expressed very often by Imp. Though zai is an
imperfective aspect marker as Imp, the use of zai in Chinese is much more limited than
that of Imp in French. Zai is a progressive marker and it only indicates ongoing
situations. It can be used to express background for punctual events, but its use is not
related to habitual situations. The use of -le is not directly associated with habitual past,
but generally speaking, it is not used with habitual events. While in the French
compositions, the first topic was intended to elicit PC, the second topic to elicit Imp and
the third one a combination of both; this distribution is not expected in the Chinese
compositions. In the Chinese compositions, -le was expected to be used most often in the
first composition because it requires many occurrences of past events, less uses were
expected in the third composition and composition two was expected to elicit the least
number of uses of-le. The use of zai was not expected to differ across different topics.
The discussion above illustrates one of the benefits of using compositions as a data
eliciting method: the content changes by topic which gives the researcher insights on the
influence of different contexts.
Written narrative is the most frequently used data elicitation method in the studies
on the acquisition of tense-aspect morphology. This method is the least influenced by the
empirical design of a study. It provides contexts that are framed in the past and offers
faithful record of learners' production. Unlike film retell which tends to elicit more
sequenced actions happening in the foreground, carefully controlled composition topics
like the ones in the present study assure a balance of predicates involved. In sum,
compositions require the least effort in the design yet provide a large quantity of material
Bardovi-Harlig (2000) highlights the weaknesses of using learner-controlled
narratives compared to cloze passages. First, it is easier in cloze passages to compare
across learners and to contrast learners' with native speakers' data because of the unified
data format. Second, the distribution of verb morphology across semantic verb classes
can be carefully controlled in cloze passages to provide a balance among different
classes. Cloze passages are a good alternative to compositions, where learners' limited
vocabulary sometimes prevents them from producing an analyzable amount of all four
verb classes. Bardovi-Harlig and Reynolds (1995) observed that the tokens of states in
interlanguage are typically dominated by the verbs to be and to have and their
equivalents. Bergstrom (1995 cited in Bardovi-Harlig, 2000) used both a cloze test and a
silent film retell in her study and found that only in the cloze passages did learners' PC
distinguish achievements from accomplishments and activities. The cloze passages give
researchers a broader sample of predicates to examine. The spread of perfective and
imperfective to their last stages are better observed with cloze passage data (Bardovi-
Harlig, 2000). For this reason, a cloze test in French and a written editing task in Chinese
were incorporated into the present study.
The format of the cloze test for French is a standard one: in the 5 passages, 37
verbs were taken out; only their infinitives were given in the parentheses. The students
were asked to provide the correct past tense of those verbs. The design of the cloze
passages proved to be a challenge. In order to have authentic materials that fit the
students' level, the book Le Petit Nicolas was chosen. The problem of using Le Petit
Nicolas is that the story is told through the voice of a little boy, so the use of PC is very
dominant. For the use of both PC and Imp, the associations between grammatical and
lexical aspects are very prototypical, namely, they are mainly between perfective and
telic verbs and between imperfective and atelic verbs. Especially lacking in the book is
the combination of telic verbs with Imp. Though there are many state verbs marked with
PC (which is not a prototypical association), when the test was given to native speakers,
they provided Imp with most of the state verbs in the places where PC was used in the
book. Changing the individual sentences did not result in more atelic verbs with PC.
What is worse, it made those sentences awkward, as judged by native speaker
consultants. An additional passage (passage 4) had to be composed by the author, with
help from a native French speaker, in order to intentionally provide more telic verbs
marked with imperfective grammatical aspect. Adding this passage helped reach the goal
in designing the cloze test, which was to have a relative balance between the use of Imp
and PC within every verb lexical class.
The Chinese written editing passage was in a completely different format. As
mentioned in Chapter Two, Chinese verbs do not conjugate according to person, tense,
mood, or voice; not every verb should be marked for aspect and the meanings among the
markers are relatively distinct (compared to French); therefore, it is not reasonable to give
a cloze test where the verbs are provided and blanks given where the verb is supposed to
be. The trick in the acquisition of the Chinese aspectual system is more about knowing
where and when to put a marker, not which marker to use. For this reason, the Chinese
task equivalent to the French cloze task was composed of a passage with no blanks and
was more accurately termed a written editing task. The aspect markers in the passage
were deleted and the students were asked to provide the markers -le, zai and -guo
wherever they thought them appropriate. -Zhe was not included in the passage because
the written editing task was originally intended for both second- and first-year students
and the latter had not learned -zhe. The format was adapted from Duff and Li (2002),
who only tested the use of -le with their written editing passage. In both studies, the
written editing task seemed to pose quite a challenge to learners.
The present study did not include any oral data because of the following two
concerns: first, oral data takes much more time to transcribe and to analyze; second, there
are difficulties involved in the analysis of French oral data. In tense-aspect morphology,
it is crucial to transcribe the verb endings correctly, but this is a challenge in French. As
mentioned in Dietrich et al. (1995), there is a problem of segmenting and categorizing the
verbal compounds in oral French. Where the suffix is concerned, /e/ may be the present
tense form of a regular er verb in second person plural or singular; the infinitive, the
past participle, or the Imparfait for all persons but 1st and 2nd plural. The verb parler is
used here to illustrate this point. All the following words are pronounced [parle]:
parlez second person plural or singular in present indicative
parle past participle
parlais first person singular in Imp
parlais second person singular in Imp
parlait third person singular in Imp
parlaient third person plural in Imp
The phoneme /e/ is really problematic in the auxiliary, as the same phoneme /e/ can
be one of three possibilities: the auxiliary avoir in first person singular, or the auxiliary
&tre in the second or the third person singular.
Data Collection Procedures
Data collection was done during normal class periods. After getting permission
from the instructors, the investigator went into the classrooms to recruit volunteers and
conducted the research. The investigator went into the classrooms three times, with one
week separating each data collection. The first time, the students filled out the language
background information sheet and completed the comprehension task; the second time,
they wrote compositions number one and two; and the third time, the students wrote the
third composition and did the cloze test. Each time, the students were told to take as
much time as they needed as long as they finished all the required tasks in one class
period; this allows approximately 25 minutes for each task. All the tasks were given at
the beginning of each data collection, so the students could decide for themselves when
to move on to the next task.
For the Chinese section, the data collection was conducted over five sessions, each
one week apart from the other, in the classroom. The difference of data collection times
between the French section and the Chinese section was entirely due to the schedule that
the investigator had worked out with the instructor to fit the schedule of his/her
classroom. The L2 Chinese students completed the language background information
sheet and the comprehension task during the first data collection, and one composition on
each of the following three weeks before completing the last task, which was the written
editing passage. Students were given about 25 minutes for each task.
Once the data collection was finished, the data were analyzed and input into
computer spreadsheets. Four native speakers of French and two native speakers of
Chinese helped provide answers to the comprehension, the cloze test in French and the
written editing task in Chinese, as well as to analyze the compositions.
For the comprehension task, each answer was recorded and coded as either correct
or incorrect according to the answers obtained from native speakers.
In each one of the French compositions, past contexts were identified. A past
context is a predicate that actually happens in the past, or a predicate that is marked with
past morphology by the student even if past tense is not required in that context. The first
case is called an obligatory context while the latter is an example of overuse of past
morphology by the learner. The combination of the two cases is called relevant contexts
(Camps, 2002). An example of an obligatory context will be: Je suis rentre chez moi le
weekend dernier (I went home last weekend) when the situation "going home" actually
happened last weekend. It is counted as an obligatory past context no matter whether the
learner used the right conjugation of the verb rentrer or not. An example of an
overgeneration would be using j'aimais lefootball (I loved soccer) to express "I love
The reason for establishing relevant past contexts is to take into consideration not
only the obligatory past contexts but also the cases of overgeneralization. If we count
only the obligatory contexts, the picture of accuracy is not complete. Overgeneralization
must be taken into account. For example, if a learner uses PC in all obligatory contexts,
the accuracy rate would be 100% if we count only obligatory contexts, but we cannot say
that the use of PC of this learner is perfect, because he might very possibly use PC when
it is not required as well. The present study calculates the accuracy rate in terms of
correct uses over all the relevant contexts, which include both the obligatory contexts and
After identifying a past context, the appropriateness of the verb morphology was
analyzed. "Appropriateness" here does not mean the accuracy of form, but that of the use.
For example, if a student uses 'ai vi forj 'ai vu (I saw), it is counted as correct because it
is a PC used in a context where PC is required. But if a student uses 'ai &td 'I was' (PC)
in the context where 'dtais 'I was'(Imp) is required, even though the form is correct, it is
registered as an incorrect use because it does not fit the context. The tense the student
used and the tense the context required were both recorded. Each verb used then was
categorized into different inherent semantic lexical aspects.
For the Chinese compositions, past relevant contexts were identified as in the
French section; a marker was counted correct if both its form (the correct choice) and
place were right. Overgeneralization was also considered when a learner provided a
marker not needed in the context.
The analysis of accuracy of the French compositions was done by two native
speakers of French along with the author. The classification of the verbs was done by the
author only; when there were uncertainties, the author's advisors were consulted. The
author and another native speaker of Chinese who is also a graduate student in linguistics
analyzed the Chinese data together. When there were discrepancies, discussions were
conducted until agreement was reached.
As Andersen and Shirai (1995) pointed out, the operational tests according to
which the verbs are classified differ in various studies; this might contribute to different
findings. The four-way verb classification, which is the analytic framework in widest use
today, was used here, based on Vendler (1967). The semantic features mentioned in
Chapter Two (Table 2.3) and repeated here, were used for the classification of the verbs
into four categories (states, activities, accomplishments and achievements) in the present
The operational tests for the present study follow those from Andersen and Shirai
(1995, p. 749, see appendix G) and Salaberry (1998). Salaberry (1998) adopted tests
specifically designed for French by Bergstrom (1995 cited in Salaberry, 1998). "Etre en
train de (to be in the process of)" is used to identify activities and accomplishments
because they can combine with the phrase, whereas states and achievements fail the test.
"En x minutes (in x minutes)" is used to distinguish accomplishments from all other
Table 4.2 Semantic features used in the verb classification
Punctual Telic Dynamic
Accomplishment -+ +
Achievement + + +
Table 4.3 Test used to determine lexical aspect for French (Bergstr6m, 1995 cited in
Salaberry, 1998, p. 519)
States Activities Accomplishments achievements
Etre en train de Yes Yes *
En x minutes Yes *
From Bergstr6m, 1995 cited in Salaberry, 1998, p. 519
Asterisks indicate that the verb fails to classify as a member of that category.
For the French cloze passage and the Chinese written editing task, four French
native speakers and two Chinese native speakers served as controls. Three native
speakers of French gave exactly the same answers and only one speaker had two different
answers out of the thirty-seven items. The two Chinese native speakers had exactly the
same answers for the obligatory uses in the Chinese cloze test, but had differences
concerning the optional uses. The differences were discussed among the native controls
and with the author until agreement was reached. Though the French cloze test was from
an authentic French text, native speaker consultants' answers were used as correct
answers when they differed from the original text.
As we discussed in Chapter Four, the tasks performed by the learners of both
Chinese and French in the present study included a comprehension task and three
compositions. In addition, there was one cloze test for the 25 French learners and a
written editing task for the 20 Chinese learners. The results from each task completed by
the learners are presented in this chapter.
There were four sentences in the French comprehension task and 20 verbs were
underlined. Students were asked to choose the time reference of those underlined verbs.
Among those 20 verbs, there were 10 used in the past, 7 in the present and 3 in the future.
The results show that the learners had a very good comprehension of time reference
in French. The average accuracy rate was 86.4% for all twenty items. Seven of the items
had an accuracy rate of more than 90%, ten of them between 80% and 90%, only three of
them less than 80%. The distribution of the ten verbs used in the past (in either Imp or
PC) over the four aspectual classes are analyzed in the following table.
Table 5.1 Accuracy of the French comprehension over the four aspectual classes
STA ACT ACC ACH
Imp 84 86 0 0
PC 92 86 96 94
Note: percentage of Imp (or PC) in each aspectual class is calculated by averaging the percentages in the
use of Imp (or PC) of all the sentences that contain predicates in that class.
We can see from table 5.1 that there was no influence of lexical aspect on the
accuracy in the comprehension of PC. The non-prototypical association, namely, the
combination between states and PC (item number 17) did not yield a lower accuracy rate
(92%). There were data available only for states and activities concerning the use of Imp,
and these two categories did not show a difference in accuracy.
Twenty-five learners completed all three compositions, generating a total of 1165
tokens. The tokens represented relevant contexts, which included both obligatory
contexts and overgeneralizations in the use of past tenses. The average number of tokens
in the learners' production was 15.5 per composition, with a range from 1 to 291. There
were 371 states, 416 activities, 355 achievements and only 23 instances of
accomplishments. Compared to the other three categories, accomplishments occupied a
very small portion (1.97%) of the 1165 tokens and were all marked by PC.
Table 5.2 Verb forms used in the three French compositions and their distribution over
four aspectual classes (percentages (raw frequencies)
Verb form STA ACT ACC ACH Total
PC 23.7 67.8 (282) 100 (23) 76.9 (273) 57.2 (666)
Imp 66(245) 27.2 (113) 0 18.6 (66) 36.4 (424)
Pres 9.4 (35) 4.6 (19) 0 2.5 (9) (63)
Plus que parfait 0 0.2(1) 0 1.1(4) (5)
Infinitive 0 0.2 (1) 0 0 (1)
Others 0.8 (3) 0 0 0.8 (3) (6)
Total 371 416 23 355 1165
Pres = present tense, Others = future, subjunctive, conditional and unidentifiable forms
1 In the composition that had only one token, the student used one sentence to talk about what the family
did during the break, then went on talking about what they normally do on vacations, using the present
States were marked mostly by Imp (66%) and both activities and achievements
were marked mostly by PC (67.8% and 76.9%). As in the cloze data, the percentage of
verbs marked by Imp decreased from states to activities and then to achievements, while
the percentage of verbs marked by PC increased in the same direction except for
accomplishments. States patterned differently from other verb categories in both the use
of PC and Imp, indicating an influence of dynamicity. In other words, the state and non-
state distinction seemed to play a role for learners when they chose tense-aspect markers.
We can also see from the table that Imp was used in only three of the four lexical
aspectual categories (states, activities, and achievements) while PC was used in all four
To determine if the use of Imp or PC was concentrated in only a few lexical items,
it is important to know how many individual verbs were used. Tokens may not give a
complete picture, as there may be a large number of tokens generated from a small
number of different verbs. There were, in total, 108 different verbs used in past contexts
in the three compositions. The verbs that were used most often (equal to or more than 20
times) included the following 11 verbs: aimer (34), aller (172), avoir (107), &tre (187),
&tudier (23),faire (52),jouer (28), manger (33), rester (20), travailler (28), voir (25).
If we count individual verbs, not tokens, the distribution of Imp and PC over the
four verb classes is shown below:
Table 5.3 Distribution of individual verbs used in Imp and PC over the four aspectual
classes in the three French compositions (percentage (raw frequencies)
Tense STA ACT ACC ACH
PC 73.3 (11) 90.7(49) 100(8) 83 (39)
Imp 86.7(13) 61.1(33) 0 31.9(15)
Total number of 15 54 8 47
Note: numbers are individual verbs not tokens, e.g. all instances of avoirs are counted as one verb.
Table 5.3 shows that there were in total 15 different states, 54 different activities, 8
different accomplishments and 47 different achievements used by the learners in their
compositions. Some verbs were used both in Imp and PC, explaining why the number of
verbs using Imp plus the number of verbs used in PC does not equal the total number of
verbs in one verb class. The most frequently used states, such as avoir, &tre, andpouvoir
can be easily used with both tenses, though the learners' tendency is to use Imp.
The distribution in Table 5.3 looks slightly different from that in Table 5.2. There
was a good balance within states between Imp and PC because the number of different
states used for Imp and PC was about the same (13 vs. 11 out of a total of 15 states). The
greater difference of percentage of the use of states between Imp and PC reflected in
Table 5.2 than in Table 5.3 is because these verbs were used more frequently (repeatedly)
with Imp. This new piece of information does not disprove the fact that states are more
often marked by Imp in learners' production; what it shows is that learners have a limited
vocabulary when it comes to states (15 verbs out of 371 tokens). As in the previous table,
accomplishments were marked only by PC. The marking of activities and achievements
was still concentrated within PC.
Table 5.4 Accuracy rate for Imp vs. PC in the three French compositions
Tense Correct Under use Over use Accuracy rate Accuracy rate in
uses in OC in OC RC
Imp 383 219 41 63.6% 59.6% (383/643)
PC 483 21 210 95.8% 67.6% (483/714)
OC = obligatory contexts, RC = relevant contexts
As shown in previous studies (Bardovi-Harlig & Bergstrom, 1996; Harley, 1989,
1992; Kaplan, 1987, Kihlstedt, 2002; Salaberry, 1998), the accuracy rate of Imp is lower
than that of PC. Table 5.4 shows that the learners used PC more accurately than Imp in
obligatory contexts as well as in relevant contexts. The underuse of Imp was a problem
among learners (219 out of 643 relevant contexts). The phenomenon in PC was the
opposite since oversuppliance (210 out of 714 relevant contexts) was more of a problem
than underuse. This difference suggests that learners were more comfortable with PC
than with Imp, and that when there was doubt about which tense to use, they often chose
PC over Imp. If we were to look only at obligatory contexts, the accuracy of PC in
learners' production would appear very high (95.8%), but because of the oversuppliance,
the accuracy in relevant contexts was only 67.6%.
Table 5.5 Accuracy of Imp and PC over the four lexical aspectual classes in the three
French compositions (percentage (appropriate uses/relevant contexts))
Tense STA ACT ACC ACH
Imp 77.2 (237/307) 49.1 (91/189) N/A 53.4 (55/103)
PC 26.5 (26/ 98) 69.7 (202/290) 100 (23/23) 82.3 (232/282)
Accomplishments constituted a particular case in the compositions since they were
marked only by PC. Besides accomplishments, all other classes confirmed the Prototype
Hypothesis. States were more accurately used with Imp, achievement verbs more
accurately used with PC. For the use of PC, the accuracy rate increased from states,
where it was the lowest, to activities, then to achievements. For the use of Imp, states had
the highest accuracy rate, but the accuracy rate for activities and achievements were
about the same (49.1% vs. 53.4%). States stood out in the use of PC and Imp, suggesting
again an influence of dynamicity in action.
Since the topics for the three compositions are different (see Appendix F, the first
topic was intended to elicit a series of past events, the second, habitual past, and the third,
a combination of the two) it was expected that PC would be used more in the first
composition, Imp more in the second, and a relative balance would be found in the third.
Table 5.6 Distribution of Imp and PC by composition
Imp PC Pres Total number of
Comp 1 27.3% (99) 67.2% (244) 4.7% (17) 363
Comp 2 46.6% (176) 45.8% (173) 7.1% (27) 378
Comp 3 35.1% (149) 58.7% (249) 4.5% (19) 424
Table 5.6 shows that in the first composition, PC was used much more often than
Imp (67.2% vs. 27.3%), as expected; in the second composition, however, contrary to
predictions, there was a balance between Imp and PC (46.6% vs. 45.8%) where a
predominance of Imp was expected; in the third composition, PC was used more often
than Imp (58.7% vs. 35.1%) while the expectation was to find a balance between the two.
Note that the expectations are only best guesses of the researcher based on what might be
found in target-like uses, so it is normal for the learners not to meet those expectations.
The topic did have an influence on the distribution of Imp and PC in that the use of PC
increased from topic 2 to topic 3 to topic 1 (45.8% to 58.7% to 67.2%) as expected, and
the use of Imp decreased in the same direction (46.6% to 35.1% to 27.3%). In other
words, the ratio of Imp over PC moved more toward a balance from topic 2 to topic 3 to
Tables 5.7, 5.8 and 5.9 show the distribution of Imp and PC over the four lexical
classes in the three compositions. The results indicate that the same trends found in all
three compositions combined together were also found in each of them individually.
Except for the use of Imp in composition 2, where the marking was fairly evenly
distributed between activities and achievements (31.7% vs. 34.6%), the percentage of
verbs marked by Imp decreased from states to activities and then to achievements; the
percentage of verbs marked by PC decreased from achievements to activities and then to
states. The trend was clearer in compositions 1 and 3 than in composition 2, because
composition 2 required more Imp than composition 1 and 3, and thus there was more of a
balance between Imp and PC. State verbs stood out as a category of verbs that patterned
differently from all other categories, again indicating the influence of dynamicity.
Table 5.7 Distribution of PC and Imp over the four aspectual classes in French
composition 1 (percentage (raw frequencies))
STA ACT ACC ACH Total
Imp 67(65) 18.8 (25) 0(0) 7.4 (9) 99
PC 27.8(27) 75.2(100) 100(11) 86.9(106) 244
Total 97 133 11 122 343
Table 5.8 Distribution of PC and Imp over the four aspectual classes in French
composition 2 (percentage (raw frequencies))
STA ACT ACC ACH Total
Imp 62.5(110) 31.7(38) 0(0) 34.6 (28) 176
PC 25(44) 64.1 (77) 100(1) 62.9(51) 173
Total 176 120 1 81 378
Table 5.9 Distribution of PC and Imp over four aspectual classes in French composition 3
(percentage (raw frequencies))
STA ACT ACC ACH Total
Imp 71.4(70) 30.7(50) 0(0) 19.1 (29) 149
PC 17.3(17) 64.4(105) 100(11) 76.3(116) 249
Total 98 163 11 152 424
The accuracy in the use of Imp and PC, as reflected in Tables 5.10-5.12, followed
the same patterns over the four verb classes as those observed in their distribution. But
the trend was less clear in composition 2. Learners did not seem to be aware of the
association between Imp and habitual past and this caused the relatively low accuracy
rate in Imp in composition 2. There were many cases of underuse of Imp in three of the
categories (states, activities and achievements), but many cases of oversuppliance of PC.
Table 5.10 Accuracy of use of PC and Imp over the four aspectual classes in French
composition 1 (appropriate use/relevant contexts = percentage)
STA ACT ACC ACH
Imp 75 (60/80) 26 (7/27) N/A 18.2 (2/11)
PC 45.5 (15/33) 80.3 (98/122) 100(11/11) 86 (104/121)
Table 5.11 Accuracy of use of PC and Imp over the four aspectual classes in French
composition 2 (appropriate use/relevant contexts = percentage)
STA ACT ACC ACH
Imp 66.3 (108/163) 39.2 (38/97) N/A 52 (27/53)
PC 21.3 (10/47) 28.6 (22/77) 100 (1/1) 51 (27/53)
Table 5.12 Accuracy of use of PC and Imp over the four aspectual classes in French
composition 3 (appropriate use/relevant contexts = percentage)
STA ACT ACC ACH
Imp 72.6 6(9/95) 60.5 (46/76) N/A 60.5 (26/43)
PC 5.5 (1/18) 73.9 (82/111) 100(11/11) 83.5 (101/121)
All the analyses above treated all the learners as a group. In the following tables,
the focus will be individual differences. When we combine all the learners together as we
did above, it appears that both Imp and PC were used with all four aspectual classes. But
this is not true for everyone. The following two tables show how Imp and PC were
combined with the four aspectual classes respectively by individual learners.
Table 5.13 Use of Imparfait with different verb classes by individual participants
Verb Class Participant ID (number of participants)
1 STA 4, 7, 12, 16, 17 (5)
2 STA, ACT 9, 13 (2)
3 STA, ACT, ACC None
4 STA, ACT, ACC, ACH None
5 STA, ACT, ACH 1, 2, 3, 5, 6, 8, 10, 11, 15, 18, 19, 20, 21,
22, 23, 24, 25 (17)
6 STA, ACH 14(1)
Table 5.13 shows that every learner in the study used Imp in their compositions,
but no one used Imp with all four aspectual classes. The latter is because there were only
23 accomplishments in all three compositions and all were used in PC. There were some
individual variations among the learners. First, among the participants belonging to the
same category (within the same row), the number of verbs used in each aspectual class is
different. For example, participants 4, 7, 12, 16 and 17 all used Imp only with states, but
participants 4 and 7 produced only one example, participant 16 produced 2 examples,
participant 12 produced 3 examples while participant 17 produced 15 examples. Second,
participants in rows 1, 2 and 5 appear to follow the pattern of the progression in the use
of Imp first with states, then with activities, then with telic verbs while participant 14 is
the only exception to this pattern.
Table 5.14 Use of Passe Compose with different verb classes by individual participants
Verb Class Participant ID (number of participants)
1 ACH None
2 ACC, ACH None
3 ACT, ACC, ACH 2, 4, 10, 20, 23 (5)
4 STA, ACT, ACC, ACH 7, 17, 19, 22, 24, 25 (6)
5 STA, ACT, ACH 5, 6, 8, 9, 11, 12, 13, 14, 15, 16, 18(11)
6 ACT, ACH 1, 3 (2)
7 STA, ACT 21 (1)
All the participants used PC in their compositions; moreover, all the participants
used PC with both telic and atelic verbs (except for participant 21). In the use of Imp,
learners moved from the most prototypical association: between states and Imp, to a less
prototypical association, namely, between activities and Imp, before finally to the least
prototypical association: between achievements and Imp. In the use of PC, no learner
relied only on the most prototypical association between achievements and PC; instead,
as mentioned above, all learners except one used both telic and atelic verbs with PC. This
piece of information alone implies that learners' use of PC is at a more advanced stage
than their use of Imp.
In the next two tables, only the participants who produced all four aspectual classes
in their compositions (it does not matter what tense the verbs were in) are included. The
rationale for these two tables is that we cannot expect a learner to use Imp or PC with
activities if s/he did not produce any activities at all.
Table 5.15 Use of Imparfait with different verb classes by participants who produced all
Verb Class Participant ID
STA 4, 7, 17
STA, ACT None
STA, ACT, ACC None
STA, ACT, ACC, ACH None
STA, ACT, ACH 2, 5, 10, 11, 19, 20, 22, 23, 24, 25
Table 5.16 Use of Passe Compose with different verb classes by participants who
produced all four classes
Verb Class Participant ID
ACC, ACH None
ACT, ACC, ACH 2, 4, 10, 20, 23
STA, ACT, ACC, ACH 7, 17, 19, 22, 24, 25
STA, ACT, ACH 5, 11
We can see from Table 5.15 that learners were either at the beginning of the
development of Imp (participants 4, 7 and 17) or at the end of the development in terms
of the distribution of Imp over the four aspectual classes. Even though they marked all 4
classes, it does not necessarily mean that they mastered the system. They may have used
very few tokens with some very specific verbs they know well.
For the use of PC, Table 5.16 indicates that learners are all at a higher level of
development than in their use of Imp.
In the French cloze test, there were five passages for a total of thirty-seven blanks
(see Appendix C). The instructions told the learners to provide the correct past tenses
(Imparfait or Passe Compose) for the verbs given in parentheses following the blanks.
The French cloze test was given to four native French speakers, three of whom
gave exactly the same answers for all the blanks. Only one person had different answers
on two of the thirty-seven items and after discussing with the other three speakers, she
agreed with their answers. The answers provided by the four native speakers and their
distribution over the four inherent lexical classes are shown in the following table:
Table 5.17 Distribution of the answers provided by native speakers over the four
aspectual classes in the French cloze test (percentage (raw frequencies))
STA ACT ACC ACH
PC 40 (4) 50 (3) 50 (2) 70.6 (12)
Imp 60 (6) 50 (3) 50 (2) 29.4 (5)
Total 10 6 4 17
PC = Passe Compose, Imp = Imparfait
The table shows that the distribution of PC and Imp over the lexical aspectual
classes was quite balanced in the cloze test. There was, however, not an exact balance,
even though great caution was taken: states were still marked more by Imp (6 Imp vs. 4
PC), while achievements were marked more by PC (12 PC vs. 5 Imp). This is because the
combination of achievements and PC and of states and Imp is a natural (prototypical)
association in native speech too.
Table 5.18 Distribution of verb forms over the four aspectual classes in the French cloze
test in learners' production (percentage (raw frequencies))
Verb form STA ACT ACC ACH
PC 25.6 (64) 54.7 (82) 58 (58) 61.6 (262)
Imp 67.6 (169) 41.3 (62) 38 (38) 31.8 (135)
Pres 1.6(4) 0.6 (1) 3 (3) 2.4 (10)
Plus que parfait 2.4 (6) 1.3 (2) 1 (1) 2.4 (10)
Infinitive 1.6(4) 0 0 0.2 (1)
Others 1.2(3) 2 (3) 0 1.7(7)
Total 250 150 100 425
Pres = present tense, Others = future, subjunctive, conditional and unidentifiable forms
The results from Table 5.18 clearly support the predictions of the Prototype
Hypothesis, namely: 1. the percentage of verbs marked by Imp dropped from states to
activities then to accomplishments and reached the lowest point at achievements; 2. the
percentage of verbs marked by PC rose from states to activities then to accomplishments
and at last to achievements.
Both activities and accomplishments were marked more by PC than by Imp. In the
use of both Imp and PC, states patterned differently than all the other verb categories,
indicating that there might be an influence of the dynamicity feature. The feature of
punctuality and telicity do not seem to have an influence on learners' choice, as there is
no pattern difference between achievements and other categories nor between telic and
The balance found in Table 5.17 is lost in Table 5.18. Across rows, the exact
balance for activities and accomplishments among the native speakers is replaced by both
categories being marked more with PC than with Imp among the learners. The gap
between the marking of Imp and PC is increased for states but decreased for
achievements. Across columns, for the marking of both Imp and PC, the effect of lexical
aspect is clearer in the learners' data than in native speakers'. There are many
manifestations of this effect in learners' data: first, in the marking of Imp, the percentage
of verbs used in the four aspectual classes clearly decreases from states to activities and
then to accomplishments before reaching the lowest point at achievements. In the
marking of PC, the percentage of verbs used in the four aspectual classes increases in the
same direction. Second, the difference between states and achievements is greater both in
the marking of Imp and PC. The fact that a relatively balanced distribution in the design
was biased in learners' data shows that the learners in the present study were indeed
influenced by lexical aspect.
Though the instructions specifically indicated that PC and Imp were the only
possible tenses to choose from, other tenses and forms, such as the present, the
pluperfect, infinitives, the conditional, the future, and some unidentifiable forms were