L 2 acquisition of English present perfect interpretations

The present study investigates the role of first language (L1), in our case Cypriot Greek (CG) or Standard Greek (SG), in the second language (L2) acquisition of English present perfect in terms of form and meaning possibilities. With respect to native speakers of CG in particular, the primary goal is to determine whether transfer from the mother-tongue, in which present perfect has only a resultative reading and simple past a resultative, an existential or a definite reading, influences the acquisition of the English present perfect. It is assumed that L2 acquisition involves establishing connections between the semantic properties/overt markers for each reading and the English present perfect. Diagnostic tests proposed by Agouraki (2006) are employed in this study, based on the (in)compatibility of certain types of adverbial markers with the existential reading and the resultative reading, respectively, as well as on the distinct semantic properties of the two readings. Almost 400 participants took part in this research. The results show that there is a certain effect of L1 on the L2 acquisition of English present perfect by CGand SG-speaking pupils, which is argued to be mainly due to the different patterns of meanings and forms in CG, SG and English.


Introduction
This paper reports on a study which deals with the second language (L2) acquisition of English by Greek Cypriot pupils, whose first language (L1) is the local variety of Greek, namely Cypriot Greek (CG).The possible transfer effect from L1 into L2 is investigated, focusing mainly on the acquisition of the present perfect.
This study aims to contribute to the general discussion in L2 acquisition research about the role of L1 along the way.Three major hypotheses about the role of Universal Grammar (UG) in L2 acquisition have been entertained over the past three decades.The first hypothesis holds that UG can be fully accessed, or, at least, that there can be direct access to UG, which means that L2 grammars are constrained by UG principles the same way as L1 grammars (see, among many others, Mazurkewich 1984;White 1989White , 2003;;Schwartz 1993Schwartz , 2003;;Eubank 1993Eubank /1994;;Schwartz & Sprouse 1996;Slabakova 2001Slabakova , 2005)).The second hypothesis assumes that UG can be partially accessed, or that there is indirect access: UG principles can be available in L2 only through L1 and parameters are either transferred from L1 or reset by UG (see White 1985;Schwartz 1987).According to the third hypothesis, UG is not accessed at all: UG principles are not available to L2 learners; instead, general problem-solving procedures are used (see Clahsen & Muysken 1986;Schachter 1988;Epstein, Flynn & Martohardjono 1996).
If the first hypothesis, i.e. the so-called Full Access Hypothesis, is the right one, and both L1 and L2 grammars are constrained by UG, transfer phenomena can take place only at certain stages of second language development.There is no unanimous opinion, however, whether learners use transfer at the later or early stages of L2 acquisition (see Pfaff 1979;Stölting 1980).A learner can also change strategies depending on the stage of L2 acquisition: initially transfer can be used, then L1 transfer together with L2 yields overgeneralization and simplification, and in the end only the latter pertains (see Andersen 1978).Typological distance and markedness should be also taken into consideration.According to the markedness theory, there are unmarked linguistic structures or items that are common or well distributed in the world's languages, and marked items that are the least distributed.Unmarked items are more easily acquired than marked ones (e.g., Wode 1977;Zobl 1979b;Eckmann, Moravcsik & Wirth 1986).Application of the transfer strategy also depends on the level of L2 knowledge on the part of the learner -the more proficient he or she is in L2, the less transfer from L1 into L2 is observed (see Andersen 1978;Meisel 1983).
The English present perfect can be described as both a category of tense and a category of aspect (see Comrie 1976): it is a secondary, non-deictic, analytically constructed tense.Secondary tenses consist of perfect and continuous forms.They relate an event not to a specific moment of occurrence as primary tenses do (e.g.present simple), but in relation to some other event.The present perfect is formed with the auxiliary verb have and the past participial form of a lexical verb.It is a compound tense, as it combines past and present: "Perfect combines two points, recognizing a situation as a single whole entity" (p.16), it indicates "continuing present relevance of a past situation, expressing the time of the state resulting from a prior situation and the time of that prior situation" (p.52).It is not compatible with past time adjuncts, such as 'yesterday ', 'last Friday', 'in 1998'.According to Huddleston & Pullum (2002: 141-146), there are four major uses of the present perfect: the continuative present perfect, which describes an event that started in the past and continues in the present (e.g.'Peter has lived in Cyprus ever since.'), the experiential/existential present perfect, which describes the situations that occurred within the period up to now (e.g.'He has been to Paris twice.'), the resultative present perfect, which shows that the result of the event/action continues in the present (e.g.'John has closed the door.'),and the perfect of the recent past, which describes a past situation that has relevance to the present, and is close in time to now (e.g.'She has just finished her dinner.').
With respect to our research, Greek Cypriot pupils would be expected to transfer from their mother-tongue (CG) into English or their acquisition of English would not be influenced at all by their L1.Transfer from L1 into L2 might be partial.If a transfer phenomenon takes place, it would be interesting to know at what stage(s) it occurs.As our area of interest is the present perfect, it is necessary to clarify the major differences between the formation and interpretation of the perfect in English and in CG, which can possibly lead to a learning problem and transfer from L1 into L2.
The paper is organized as follows.Section 2 continues the discussion on the readings associated with the perfect.Section 3 describes the differences between present perfect and simple past, and looks more closely at the existential reading and the resultative reading associated with the present perfect.Section 4 presents the methodology of the experimental study.The results of the study, implications for further research, as well as a short discussion of the role of Standard Greek in L2 acquisition of English present perfect by Cypriot Greek speakers are provided in Section 5. Finally, Section 6 raises the more general question of how the readings associated with the perfect could be represented in Universal Grammar, and the implications for L2 learning.

Differences between existing typologies
We will start this section with a terminological point, namely that we will be using the description readings associated with the perfect, rather than readings of the perfect.This will be so because languages which lack the perfect can nonetheless express the whole range of these readings, obviously using other means.If so, in languages that distinguish between a present perfect and a past tense the readings under examination cannot, strictly speaking, be described as readings of the perfect.They are rather readings associated with the perfect, in the sense of being marked with the perfect.
In the relevant literature there are partly different typologies of the readings associated with the perfect.But, perhaps more importantly, these typologies may also differ in terms of the reading they assign to the use of the perfect in a particular sentence.To illustrate our point, it suffices to refer to two of these typologies, i.e.Iatridou et al.'s (2001) typology and Portner's (2003) typology.Iatridou et al. (2001)  (2) a. Mary has read Middlemarch.b.The Earth has been hit by giant asteroids before (and it probably will be again).c. John has been in Baltimore since yesterday.d.The Orioles have won!Portner (2003:460) goes on to remark that "we also find terms used like 'experiential perfect' and 'current relevance perfect' applying to some of the same data".He then points out that "a core experiential perfect might be (3) [: our numbering], since it says something about an experience of Mary; a good example of a current relevance perfect would be 2(b) [: our numbering], on the understanding that the potential for asteroid impacts is relevant to our decisions today." (3) Mary has eaten breakfast already.
A comparison of Iatridou et al.'s typology and Portner's typology of readings associated with the perfect reveals (a) that the typologies do not include exactly the same set of readings.And (b) that the same type of example (for instance 1(b) and 2(a)) is not necessarily assigned the same reading in both typologies.More specifically, 1(b) is taken by Iatridou et al. to illustrate the experiential/ existential reading, while similar example 2(a) is taken by Portner to illustrate the resultative reading.Also, example (3) is taken by Portner to illustrate the experiential reading, while, most probably, it would be taken by Iatridou et al. to illustrate the recent past reading or the resultative reading.At the same time, what the difference described in point (b) shows is that the common terms in the two typologies do not necessarily have the same content.More generally, the differences between the existing typologies show that the different readings associated with the perfect are not adequately defined, or rather that the differences among them are not adequately defined.In addition, it is also probably the case that the core meaning shared by all these readings associated with the perfect has not been clearly identified either.
On the basis of the characterization of examples provided by Iatridou et al. and Portner, the

An alternative typology for the readings associated with the perfect
The aim in this section is to sketch a typology of the readings associated with the perfect that can (a) identify the core meaning associated with the perfect in all the distinct readings associated with the perfect, and isolate this core meaning of the perfect in all the readings associated with the perfect, (b) account for the observed readings, and (c) address the issues raised with respect to the two typologies presented in the previous section, as well as the cross-linguistic differences in the readings associated with the perfect.As known, languages that distinguish between a present perfect and a past tense do not all have the full range of readings associated with the perfect in English.More specifically, Standard Greek (SG) lacks the continuative/ universal perfect and the hot news perfect.This obviously does not mean that SG lacks the continuative/ universal reading and the hot news reading in general; only that these two readings are not associated with the perfect in SG.The continuative/ universal reading is instead expressed with present tense, while the hot news reading is expressed with verb-initial sentences in past tense (cf.Section 6).Cross-linguistic differences in the readings associated with the perfect are generally accounted for in terms of syntactic or morphological properties of the languages in question.
Next, we present a new typology for the readings associated with the perfect.In this typology the readings associated with the perfect are reduced to independent readings, on which the meaning of the perfect, described in the literature as 'the current relevance of past event' meaning component, is overlaid (for an extended discussion of the proposed typology see Agouraki 2011).It is proposed that the independent readings on which the meaning of the perfect can be overlaid include the continuative/ universal reading, the habitual reading, the definite reading and the indefinite/ characterizing reading.Moreover, the definite reading and the indefinite/ characterizing reading may optionally additionally bear one of the following interpretations, i.e. the resultative interpretation, the quantificational interpretation or the enumerative interpretation.We will only comment on the readings that are either not self-explanatory or have not been presented as such before.

Continuative/ Universal reading:
(4) Mrs Dade's been phoning her home number all day.

Habitual reading:
(5) a.I'm 59 years old and have been watching these shows all my life.
b.I have slept more soundly since Obama became President.

Definite reading:
We have called this reading the definite reading, as opposed to the continuative reading or the habitual reading, because the tense refers to some more or less definite past time.At the same time, the definite past time is contextually relevant.And the event is more or less recent (cf.6(a) and 6(b), respectively).In that sense, the definite reading apparently covers the recent past reading and the hot news reading of other typologies.
(6) a. I've called the police and they are on their way.b.Kurt Wallander has made the dream of his life come true.He has bought a house by the sea.
The contextual relevance of the past event is precisely the property that differentiates the definite reading associated with the perfect from the definite reading discussed with respect to past tense.In fact, the description 'contextually relevant past time' is not unrelated to the description 'current relevance of past event', generally taken to be the core meaning of the perfect, in the sense that whenever reference is made to a contextually relevant past time there obtains at the same time current relevance of the past event described.If we take away the 'current relevance of past event' meaning component, this reading is no different than what is generally referred to as the definite reading of past tense.As far as the temporal proximity of the past event is concerned, this second characteristic of the definite reading associated with the perfect is taken to be a consequence of the interaction between the current relevance of past event and the definite reading.Some evidence for a correlation between the definiteness scale and the anteriority scale comes from the observation that the more definite the past time is, the more recent the event is.
It is proposed that the so-called resultative reading associated with the perfect (cf.( 7)) is not a distinct reading by itself but a supplementary feature on the definite reading.
Crucially, it is claimed that verb classes affect the interpretation of the definite reading, in the sense of potentially adding a semantic feature to it.More specifically, it is claimed with respect to the resultative interpretation associated with the perfect that it is unaccusative predicates or predicates with an unaccusative component that yield a resultative interpretation of the definite reading.
The Indefinite/ Characterizing reading: The indefinite/ characterizing reading is in essence no different than what is standardly referred to as the experiential reading, taken in what we see as its core case, i.e. with no number of times adverbial (cf.(8)).
In contrast with the definite reading, the indefinite/ characterizing reading does not involve a contextually relevant past time.It remains to be identified how the 'current relevance of past event' meaning component of the perfect interacts with the independent indefinite/ characterizing reading.It appears that this interaction yields a reading that could be described as an "about the person" reading.The human subject has had at no contextually relevant past time an experience that tells us something about his physical abilities, his character, his intellect, his aptitudes, his mentality, his way of life, or his life--style.In short, this experience tells us something about what he is as a person.Sometimes the experience is of the type that there can have been more than one occurrence of this experience, but it often does not make a difference, as far as the knowledge about the person's life is concerned, how many occurrences of this experience there have been.For instance, it does not make a difference for this reading whether one has eaten sushi one or more times.The most common, implicit, temporal expression that is compatible with the indefinite reading is the expression in my life.In the case of inanimate subjects, the information is about the properties of that entity (cf.Portner's 2(b)).The "about the author" page of books typically includes uses of the indefinite/ characterizing reading (cf.( 9)).
(9) Since her first novel, From Doon with Death, published in 1964, Ruth Rendell has won many awards, including … .Her books have been translated into twenty-five languages.

Interpretations over and above the proposed readings:
A number of factors may play a role in the instantiation of the definite or the indefinite/characterizing reading.
Another factor which results in a modification of the definite or the indefinite/ characterizing reading associated with the perfect is whether there is an adverb of numbers in the sentence.If so, a quantificational interpretation is superimposed on the existing definite or indefinite reading.
(10) a.I have already/ only called him twice.b.I have taken my theory test three times.
(11) a.He has held the World Championship three times.b.Faye Dunaway has been married twice.
In the case where the quantificational interpretation is superimposed on the indefinite/ characterizing reading the period during which there has been at least one eventuality of a particular type often amounts to a person's life.
The quantificational interpretation superimposed on a definite or indefinite reading apparently corresponds to the existential reading of the perfect in other typologies.Although, as shown in (10), the quantificational interpretation is equally compatible with the definite reading, the examples used in other typologies to illustrate the existential reading standardly involve the indefinite/ characterizing reading combined with a quantificational interpretation (cf.( 11)).The question arises as to what purpose is served by decomposing the existential reading as we did.This practice serves the general purpose of showing that the so-called readings of the perfect involve independent meaning components associated with the perfect.As shown in (12), the quantificational interpretation is not exclusive to the perfect; rather, it is compatible with the perfect.Apart from unaccusativity and adverbs of numbers, another factor that may play a role in the instantiation of the definite or the indefinite/ characterizing reading is whether the text marks a sequence of events, which yields an enumerative interpretation.The enumerative interpretation is optionally combinable with the definite reading and the indefinite/ characterizing reading (cf.( 13)).Informally speaking, the enumerative interpretation seems to constitute an appropriate answer to the question What has X done today/ since then/ in their lives.It portrays a sequence of events.
(13) Since I left school (at 14 with no qualifications) I have been engaged, had a beautiful boy called Jay, who is now 6, left his father when Jay was 3 months old, sorted stuff out on my own, had three very good jobs, right now I am working for the Football and Cricket clubs.
The proposed typology of the readings associated with the perfect differs from alternative typologies in terms of main idea, as well as in terms of coverage.(a) The main idea behind this typology is that the only meaning of the perfect is the 'current relevance of past event' meaning.Everything else in the so-called readings of the perfect is independent readings, which are combinable with the perfect.This is what has motivated the rethinking of the recent past reading, the hot news reading, the resultative reading and the existential reading.(b) An additional reading the meaning of the perfect can combine with has been identified, i.e. the habitual reading.
The second part of this paper (Sections 3-5) discusses L2 acquisition of English present perfect interpretations by native speakers of CG or native speakers of SG.So far we have argued that the generally accepted readings associated with the perfect consist of more basic component parts, and are differentiated from one another in terms of distinct associations of these component parts.In the discussion to follow, for ease of reference, we will be using the generally accepted terms resultative reading and existential reading.These correspond to the definite reading with unaccusatives and the indefinite reading with a quantificational interpretation, respectively.The discussion carried out in section 2 will again become relevant in section 6, which considers how the readings associated with the perfect are encoded in Universal Grammar.

Differences between past and perfect
According to Reichenbach (1947), the temporal location of every event (E) is specified by its relationship to the moment of speech (S), so for the past tense E is before S, for the present tense E coincides with S, and for the future tense E is after S. Reichenbach assumed reference time (R) to mediate between S and E. These relations can be coincidence (",") or precedence ("_").So, within this approach, simple past is E,R_S (the event coincides with the reference point, which precedes the moment of speech), and present perfect is E_R,S (the event precedes the reference point, which coincides with the moment of speech).Though for some tenses (e.g., simple past: E,R_S), R seems to be superfluous as R and E coincide, R is very important for the description of the complex tenses such as future perfect (e.g., S_E_R: 'Mary will have arrived in London by 6 o'clock.').The event ('Mary's arriving in London.') is placed between the moment of speech and a reference point.Iatridou et al. (2001) establish three main differences between present perfect and past tense, though both present perfect and simple past express precedence or anteriority.Firstly, simple past reference time precedes speech time (R_S) and perfect event time precedes reference time (E_R) (Reichenbach 1947;Hornstein 1990).Secondly, according to McCoard (1978), Klein (1992) and Giorgi & Pianesi (1998), simple past and present perfect are compatible with different adverbs: past-oriented for simple past (e.g., 'yesterday', 'in 1990': 'I saw this film last month.')and perfect--oriented for present perfect (e.g., 'since', 'for': 'I have not seen you since 1991.').Thirdly, simple past and present perfect have aspectual differences: the perfect focuses on the state that follows from a prior event (Parsons 1990;Vlach 1993;Giorgi & Pianesi 1998), whereas the simple past has other aspectual properties, namely, stative (e.g., 'He loved Mary.') and non-stative -including accomplishment (e.g., 'She painted a picture.'),achievement (e.g., 'Tom won a race.')and activity (e.g., 'He jumped.')-the latter will not be pursued any further in this paper.

Existential perfect vs. perfect of result
According to Iatridou et al. (2001) and Agouraki (2006), perfect requires an interval, the so-called perfect time span: the right boundary (RB) of the perfect time span is set by tense, while the left boundary (LB) is set by perfect adverbials (e.g., 'since': 'Mary has visited England since 1991.').This is an example of the existential perfect, the RB of the perfect time span is now, the time of the utterance, and the LB of the perfect time span is some time in 1991 and there is an interval from 1991 until now, of which it is true that at least once in that interval Mary visited England.
The existential perfect presupposes that there is an interval in which at least one (un)bounded eventuality occurs (Agouraki 2006).The (un)bounded eventuality has to be completed and contained in the interval (e.g., 'I have called you three times since then.' or 'I have lived in London since 1995.').The perfect of result, on the other hand, requires telic predicates and presupposes the existence of the underlying eventuality (e.g., 'I have broken my arm.').If the arm is still broken, it is the perfect of result, but if the arm has healed, it is the existential perfect.
Many researchers believe that the perfect of result is not an independent category but a subcategory of the existential perfect.Brugger (1997) and Kratzer (2003), though, suggest that the perfect of result is an independent semantic category.They follow Parsons ' (1990) analysis of an event's target state: every event that culminates (e.g., throwing a ball) has a target state (e.g., a ball being thrown onto the roof), which may or not last for a long time, and a resultant state (e.g., someone's having thrown the ball onto the roof) that holds forever after.According to Kratzer (2003), the perfect of result is encoded in the meaning of adjectival suffixes in adjectival participles.

Present perfect and past simple: SG vs. CG
Standard Greek (SG) has two forms for the present perfect: Present Perfect A and Present Perfect B. Present Perfect A is formed with the auxiliary verb eho 'have' and the perfective participle (e.g., eho akusi 'I have heard').The perfective participle does not show either agreement or tense, the temporal information comes from the auxiliary.For transitive verbs, Present Perfect B is formed with the auxiliary verb eho 'have' and a participle agreeing in phi--features with the object (e.g., eho mayiremeno 'I have cooked') or a participle with default, neuter plural agreement in the case of an understood 'typical' object (e.g.eho mayiremena 'I have cooked').For intransitive verbs, Present Perfect B is formed mainly with the auxiliary ime 'be' and a participle agreeing in phi-features with the subject (e.g., ime yenimenos 'I was born'), and occasionally with the auxiliary eho 'have' and a participle agreeing in phi--features with the subject (e.g., eho padremenos 'I have been married').Present Perfect B with ime 'be'/eho 'have' + adjectival participle can be formed only with a subclass of intransitive verbs, i.e. unaccusative verbs (Agouraki, 2006).Unaccusative verbs are intransitive and have non-agentive subjects (e.g., 'a glass dropped'; 'the door closed').Unergatives, which are also intransitive but have agentive subjects (e.g., 'laugh'; 'swim') do not form Present Perfect B. With respect to SG, Agouraki (2006) claims that Present Perfect A can have an existential reading or a resultative reading (e.g., eho diavasi afto to vivlio 'I have read this book.'),among other readings; Present Perfect B, on the other hand, can only have a resultative reading (e.g., eho vamena ta malia mu 'I have my hair dyed').Past tense can have a definite temporal reading, an indefinite reading, an existential reading or a resultative reading (e.g., diavasa to vivlio 'I read this book).
CG lacks Present Perfect A1 (Menardos 1969), but has Present Perfect B can only have a resultative reading (e.g., eho mairemena ta faya 'I have the food cooked').Past tense in CG can have the same range of interpretations as in SG, that is a definite reading, an indefinite reading, an existential reading or a resultative reading.It should be pointed out, however, that the use of Present Perfect B in CG, as in SG, is diminishing, and that nowadays the main means of expressing the resultative reading is through past tense.Which brings us to the following point.Namely, if we adopted Kratzer's (2003) analysis of the perfect of result and the contribution of the adjectival participle, could it be claimed that in Greek Present Perfect B the perfect of result is encoded in the meaning of the agreeing participle (i.e.-menos as in grafo-gramenos, 'write'-'written') and that verbs that can form adjectival participles can have a target state and a resultative reading?It seems that this claim cannot go through for the following reasons.It can only describe the resultative reading with Present Perfect B in both SG and CG, but it cannot describe the resultative reading with Present Perfect A in SG, or the resultative reading with Past Tense in both SG and CG.
English present perfect can have an existential or a resultative reading, among other readings.As already presented, the paper focuses on L2 acquisition of English existential perfect and resultative perfect.The possible, expected learning problem of CG-speaking pupils acquiring English could be the use of past tense instead of the existential present perfect.Since in CG the existential reading is one of the readings of the past tense, which is, however, obligatorily marked with specific adverbials (e.g., potte 'ever', kammian foran 'anytime', stin zoin mu 'in my life', os tora 'up to now'), it would be interesting to see which tense Greek Cypriot pupils learning English use with these adverbials in English.The also anticipated use of past tense for the resultative reading is less of a problem, as it is possible in English as well to use the past tense for the resultative reading (e.g., I didn't bring my car.).Another possible learning problem of CG-speaking pupils acquiring English could be if they over-generalized and used the present perfect in English also for the definite and the indefinite readings, with the additional use of past time adverbs with the present perfect.To sum up, we examine whether there is transfer from L1 into L2, i.e. use of past tense with the existential meaning instead of present perfect with the existential meaning or use of past tense with the resultative meaning instead of the resultative present perfect.L1 transfer will yield ungrammatical sentences in the first case, but possibly grammatical sentences in the second case.
The grammaticality judgment tests (both in Greek and in English) collected in this research are based on the diagnostic tests from Agouraki (2006).We tested if pupils' errors and deviations from the norm were due to a transfer effect, and if so, at what age pupils used the transfer strategy.To test the hypotheses (transfer vs. non-transfer, stages when transfer occurs), four groups of pupils took part in the study: Greek Cypriot pupils from government schools, Greek Cypriot pupils from private schools, Greek pupils from government schools, and a control group from a British private schoolnative speakers of English.According to the school curriculum for English, both in Cyprus and Greece, the present perfect tense is introduced for the first time in gymnasium.Consequently, some Greek and Greek Cypriot pupils had just been formally introduced to the present perfect tense (i.e.gymnasium pupils) and some had been acquainted with this tense for several years (lyceum pupils).The picture is more complicated, however, as it is often the case that students start taking English lessons at an early age, when they are in primary school.This means that gymnasium students have also been acquainted with the present perfect for a few years.

Participants
In total, 398 pupils took part in the study.All the participants were volunteers and were recruited from 17 different schools in Greece and in Cyprus: 117 pupils from seven different government schools in Cyprus, 111 pupils from three different private English-speaking schools in Cyprus, 156 pupils from six different government schools in Greece, and a control group of 14 pupils, native speakers of English, from a British private school in Cyprus.The sample was gender balanced, there was nearly the same number of female (209 pupils, 52%) and male (189 pupils, 48%) participants.The study was conducted both in urban and rural areas.The primary school pupils (5 th and 6 th grade) that took part in the research were 10-11 years old (105 pupils, 26%), the gymnasium pupils (1 st to 3 rd grade) were 12-15 years old (120 pupils, 30%) and the lyceum pupils (1 st to 3 rd grade) were 16-19 years old (173 pupils, 44%).As a point of clarification, children in Cyprus attend six grades in primary school (typically from age 6 to 12) and six grades in secondary school, the first three of which are called gymnasium (age 12 to 15) and the other three lyceum (age 15 to 18), after which they get the school--leaving diploma with which they can apply to university, for example.

Materials
The material used in this study consisted of a questionnaire and two tests for the pupils.The questionnaire had 23 questions (multiple-choice, yes-no, and open-ended).The 23 questions requested the following information (in this order): sex, age, origin, school, class, father's education, mother's education, socioeconomic status of the family, mother-tongue, grade/mark in English at school, grade/mark in Greek at school, years of learning English, knowledge of other languages, attitude towards the English culture, reasons for learning English, language difficulty, language of daily use, role of mother-tongue in the acquisition of English, friends or relatives speaking English, visiting English-speaking countries, extra lessons in tutor centre and attitude towards learning English.
The tests consisted of a sentence-picture matching task (SPM) and a cloze task.The SPM task included 27 items: for 20 items pupils needed to choose between simple past and present perfect (e.g., 'They have already started their dinner.' vs. 'They already started their dinner.'),7 items were distractors, where pupils had to choose between simple present, present continuous, simple future, and past continuous (e.g., 'She is talking on the phone right now.' vs. 'She talks on the phone right now.').The cloze task included 19 items: for 13 items pupils needed to choose between simple past and present perfect (e.g., 'Ann is so happy.She won the first prize.' vs. 'She has won the first prize.');6 items were distractors, where pupils had to choose between simple present, present continuous, simple future, and past continuous (e.g., 'She is flying to Spain tomorrow.' vs. 'She flies to Spain tomorrow.').The tests were based on the diagnostic tests by Agouraki (2006).

Procedure
Pupils were given about one hour and a half to fill in the questionnaire and complete the tests.They were assured that their performance would be kept confidential.The questionnaire and the tests were pilot-tested, tried on other groups of pupils of the same age.Some problematic items were detected and revised for the final study.

Results and discussion
It is assumed that L2 acquisition involves establishing connections between the semantic properties/overt markers for each reading, and the English present perfect.Diagnostic tests proposed by Agouraki (2006) are employed in this study, based on the (in)compatibility of certain types of adverbial markers with the existential reading and the resultative reading, respectively, as well as on the distinct semantic properties of the two readings.To check whether there is transfer from mother-tongue (from CG/SG into English), tests were offered to the pupils in which they needed to choose between simple past and present perfect in various contexts, in combination with different adverbials.

Possible vagueness with respect to the number of events
According to Iatridou et al. (2001: 200) and Agouraki (2006: 46), "the existential perfect has the interpretation of at least once, or more than one time" (e.g., 'He has been to England many times.'),while "the result reading involves only one eventuality" (e.g., 'He has gone to England.').Following one diagnostic test from Agouraki (2006), 'possible vagueness with respect to the number of events', the pupils were given sentences with an existential reading in order for them to choose between simple past and present perfect.Having in mind that Greek Cypriot pupils might choose simple past (in case of transfer from CG where simple past expresses the existential meaning).There were three examples of this diagnostic test.One such example is given in ( 14), which consists of two sentences: one in simple past tense, the other in present perfect tense; the target is the existential present perfect that presupposes the interpretation at least once, more than one time.

Temporal placement of the event
According to the 'temporal placement of the event' diagnostic test (Iatridou et al. 2001;Agouraki 2006), the eventuality in the perfect of result is obligatorily placed at the LB of the perfect time span, where the LB is set by the adverbial (e.g., 'since') and the RB is set by utterance time (cf.'He has changed his address since April.'),while in the case of the existential perfect the eventuality cannot be placed at either the LB or the RB (cf.'He has changed his address two times since April.').There were four cases of this diagnostic test.Present perfect in CG can only express the resultative reading, while simple past can have a resultative, an existential or a definite reading.In the case of positive transfer from CG into English, Greek Cypriot pupils might choose the correct variant, i.e. present perfect, and in the case of negative transfer the incorrect variant, i.e. simple past.One example is given in (15), where pupils were offered two sentences, one with present perfect and the other with simple past, where the target is the perfect of result that obligatorily places the eventuality ('finishing shopping') at the LB of the time span, while the RB is the utterance time ('now').
(15) a.My e-mail accounts have been disabled since we moved house.
b. My e-mail accounts were disabled since we moved house.

The how long-question test
The how long-question presupposes the interval that started when the eventuality was completed until the utterance time.This question is possible with a resultative reading (e.g., 'I have cooked.'),but not with an existential reading (e.g., 'I have jumped with a parachute two times since March.').There were two cases of this diagnostic test, e.g. ( 16), where pupils were asked to choose between two sentences, one in present perfect and the other one in simple past, the target being the sentence with the present perfect; the picture was intended to help the participants understand the relevance to the present time.
(16) a.How long have you been done with the trends?b.How long were you done with the trends?
Under normal circumstances, in the case of a resultative reading of the perfect, GC-speaking children would be predicted to choose present perfect (if there is positive transfer from CG, as present perfect expresses the resultative reading and cannot express the existential reading in CG), and simple past (if there is negative transfer from CG, as in CG simple past can have the definite reading, the existential reading or the resultative reading).
Methodologically, both the second ('temporal placement of the event') and the third tests (how long-question) may not seem so strong, since in CG, besides present perfect, simple past can also express the resultative meaning.To counter possible objections, we tried to have two targets in the test, both existential perfect and the perfect of result, to see whether both positive and negative transfer occur in the pupils' performance.

Interpretation of the ksana-test
According to Agouraki (2006), the use of existential perfect with ksana 'again' presupposes that there has been at least one other occurrence of the eventuality expressed in the proposition (e.g., To 1991 diavasa afto to vivlio, apo tote to eho ksanadiavasi.'I read this book in 1991, since then I have read it again.').
There were five cases of this diagnostic test.The sentences with the existential perfect were chosen and included in the test.In CG the present perfect can only express a resultative reading, while simple past can express a definite, a resultative or an existential reading.In case of transfer from CG into English, participants were expected to choose simple past.There is an example of this type of the test in ( 17), where pupils were offered two similar sentences, one with present perfect tense and the other with simple past; the target was the existential present perfect.
(17) a.Since then the scientist has done this experiment again.
b. Since then the scientist did this experiment again.

Markers of the existential perfect mehri simera 'up to today', mehri tora 'up to now'
According to Agouraki (2006), the adverbial modifier mehri tora 'up to now' and mehri simera 'up to today' are markers of the existential reading in CG.Six sentences with the existential present perfect were chosen and included in the test.Since present perfect cannot express an existential meaning in CG, while simple past can have an existential or a resultative interpretation, Greek Cypriot pupils might choose simple past (in case of transfer from CG).One example of this diagnostic tool is in ( 18), where pupils were offered two sentences, one with present perfect and the other one with simple past; the target was the existential present perfect.
(18) a.I have not caught any fish up to now.
b.I did not catch any fish up to now.
4.4.6.Additional existential markers: potte 'ever', kammian foran 'ever', stin zoin mu 'in my life' According to Agouraki (2006), there are additional markers for the existential reading, such as potte 'ever', kammian foran 'ever' and stin zoin mu 'in my life'.We used ten sentences with the existential present perfect in the test.The existential reading can be expressed in CG only with simple past, so participants might choose simple past instead of present perfect (in case of transfer from CG into English).One example of these sentences is in (19), where pupils were asked to choose between two sentences: one in present perfect and the other one in simple past; the target was the existential present perfect.
(19) a. Have you ever tried skiing?b.Did you ever try skiing?

Test performance
The "easiest" test for all types of schools was the 'temporal placement of the event' test (section 4.4.2.).As for the most "difficult" test, some differences could be observed.For government schools in both Cyprus and Greece, the 'possible vagueness wrt the number of events' test was the most difficult one (section 4.4.1.).For private schools in Cyprus, pupils performed worst on the fifth test (section 4.4.5.).For British pupils in Cyprus, the most difficult task was the sixth test (section 4.4.6.).Table 1 presents the production of all the groups in each one of the tests.It can be seen that Cypriot government school pupils tend to use more Simple Past than the other groups, with the exception of the 2 nd test: temporal placement of the event.This means that they probably transfer from their L1 Cypriot Greek and use Simple Past instead of Present Perfect.Unsurprisingly, the British private school pupils showed the best test performance (89.26% correct answers).Cypriot private school pupils were the second best with 70.56%, then followed Greek government school pupils (67.19%) and finally the pupils from Cypriot government schools (64.14%).However, the difference between Cypriot government school pupils and Greek government school pupils is very narrow, so that it can be assumed that this difference might be due to transfer from CG into English or to the difference in the amount of L2 input.Maybe the quality and focus of teaching in different educational settings -Greece and Cyprus, private and government schools -played a role, too.The results are presented in Table 2 and Figure 1.According to a two-sample t-test, there is generally a statistically significant difference between the pupils' test performance of all types of schools in all the tests: But if both Cypriot government school pupils and Cypriot private school pupils are taken as a single group and compared with the group of Greek government pupils, there is no difference in the performance of Greek Cypriot and Mainland Greek pupils.This leaves us with a lower possibility to assume that transfer from CG plays a crucial role in the acquisition of English present perfect by Greek Cypriot pupils.

School type division
Comparing the test performance of the participants from different schools (Cypriot government and private, Greek government and British private) and of different age (pupils of the primary school, gymnasium and lyceum), it can be seen that for Cypriot government schools, the best total test results were achieved by lyceum pupils (16-19 years old), for Greek government schools the best results were achieved by lyceum pupils (16-19 years old), and for Cypriot private school the best total test performance was achieved by gymnasium pupils; see Table 5 and Figure 3  For Cypriot government schools, there is a statistically significant difference between primary school and gymnasium, as well as between primary school and lyceum, but there is no statistically significant difference between gymnasium and lyceum.The same is true for Greek government schools: there is a statistically significant difference between primary school and gymnasium, as well as between primary school and lyceum, but there is no statistically significant difference between lyceum and gymnasium.For Cypriot private schools, there is a statistically significant difference among primary school, gymnasium, and lyceum.An interesting trend is that for Cyprus government school and Greek government school there is an improvement of the pupils' performance, from primary to gymnasium to lyceum, while the pattern is not the same for Cypriot private schools' production: there is an improvement from primary to gymnasium, but from gymnasium to lyceum there is a decrease of target present perfect production.This can be explained by differences in school curriculum, grammar syllabus, as present perfect is mainly introduced and practiced at gymnasium level, or by a possible peer influence from non-private schools, home and non-school environment, where only CG is spoken.

Conclusions
The study investigated the role of L2 input in the acquisition of the resultative vs. the existential interpretation of English present perfect in different educational (private vs. government) and geographical (Greece vs. Cyprus) settings.The results suggest that the amount of L2 input plays a crucial role in L2 present perfect acquisition, as a statistically significant difference between the test performance of government and private school pupils in Cyprus was revealed.The language of teaching in Cyprus private schools is English, so their pupils are exposed more to English than the pupils of the government schools, where all the subjects are taught in Greek and they have English lessons two or three times a week only.
Another finding is that there is a certain effect of L1 on L2 present perfect acquisition, as a statistically significant difference was found between CG pupils and SG pupils of government schools in their test performance, and this difference is mainly due to the difference in available meanings, rather than to a difference in the form of the present perfect in CG and SG.There are not so many secondary English-speaking private schools in Greece, which is why it was difficult to find participants to match the group of Cypriot private school pupils.On this basis, the data from Cyprus concerning the private sector of education cannot be compared with relevant data from Greece, though there is a statistically significant difference between government school pupils' test production in Greece and in Cyprus.
The age of the participants and their levels of education and age of exposure to L2 can be seen to influence their test production; the older the learners were and the more years of exposure to L2 they had, the better test production they showed (lyceum and gymnasium pupils performed better than primary school pupils, except for the Cypriot private school group).Overall, the findings of the study can be taken to support the Full Transfer/Full Access Hypothesis (Schwartz & Sprouse, 1994, 1996;Schwartz, 2003;White, 2003;Slabakova, 2001Slabakova, , 2005)), according to which there would be transfer from L1 (CG) into L2 (English) concerning the acquisition of the present perfect, and transfer decreases when the level of L2 proficiency and length of exposure increases -though we hasten to add that the claim may not be as strong, since there is no data from Greek private schools.
Some other issues should be explored as well before reaching a final, fast conclusion.In particular, there are three areas that call for additional study.These include (i) the frequency of present perfect (resultative and existential) and simple past in both CG and SG, (ii) the age when the young learners differentiate between simple past and present perfect (existential vs. resultative) in their native languages (CG and SG), and (iii) the school curriculum in Cyprus and Greece, in private and government educational settings with the focus on the teaching of tenses (simple past and present perfect).Lastly, further research within the framework of the Interpretability Hypothesis (Tsimpli, 2003;Hawkins & Hattori, 2006), relevant to the syntax-semantics and syntax-discourse interfaces, is needed for a deeper understanding of whether we are dealing here with transfer from L1 or difficulties in interpretability in L2.

The role of Standard Greek in L2 acquisition of English present perfect by Greek Cypriot speakers
An important parameter that most probably plays a role in how we read off the study's results has to do with the fact that all Greek Cypriot students who attend Greek-speaking schools have already acquired SG before they acquire English, SG being the official dialect and the dialect of education in Cyprus.Why is this crucial?Because, contrary to CG, SG has Present Perfect A, and an existential reading associated with it, among other readings.So the first task of Greek Cypriot students is to acquire the readings associated with the perfect in SG, which, crucially, include the existential reading.It is only later that they learn the readings associated with the perfect in English.The obvious question is whether acquisition of the readings associated with the perfect in English is mediated through the acquisition of the readings associated with the perfect in SG.
In fact, it probably does not make a huge difference whether Greek Cypriot students attend a Greek-speaking school or an English-speaking school, as (a) Greek Cypriots are exposed to SG before they go to school, crucially from a very early age, as SG is heard a lot in Cyprus (TV, radio, SG speakers around, and Cypriots visiting Greece a lot) and (b) we doubt that, apart from the form of Present Perfect A, Greek Cypriot students attending Greek-speaking schools are explicitly taught anything about the different readings associated with Present Perfect A. All their grammar book says about the present perfect is that it is another 'present' tense.Thus, it is mostly a question of how well they acquire the second dialect that they are 'exposed' to from an early age.
Related to the "SG as an intermediary" parameter is the observation that the set of readings associated with the perfect in SG differs from the set of readings associated with the perfect in English.Namely, SG lacks the universal reading and the "hot news" reading with the perfect.
A second issue related to the "SG as an intermediary" parameter is how Greek Cypriot students learn the set of readings associated with the perfect in SG versus how they learn the set of readings associated with the perfect in English.Crucially, the mode of acquiring SG (the set of readings associated with the perfect included) differs from the mode of acquiring English (the set of readings associated with the perfect included).The first acquisition process takes place inside and outside the school environment but, crucially, does not involve formal instruction on the syntactic level.It is a general fact that the syntactic differences between CG and SG are not discussed in Cypriot schools.Which leaves open the question if and how many Greek Cypriot students know the set of readings associated with the perfect in SG.The only piece of information they certainly have is the form of the present perfect in SG.
With respect to CG having Present Perfect B, one can observe the following.(a) Present Perfect B for unaccusatives, i.e. the auxiliary be plus a -menos participle agreeing in phi-features with the subject, is most probably analyzed by Greek Cypriot students as a copula construction, and not as present perfect.If so, there is reason to doubt that it would give positive transfer for L2 acquisition of the English present perfect.(b) It is only in the name present perfect or in terms of the resultative interpretation, if one had explicit grammatical knowledge, that a resemblance between the CG present perfect and the English present perfect would be recognized by Greek Cypriot students.But this is highly unlikely, as CG, the native dialect of Greek Cypriots, is not taught/ described in schools in Cyprus.(c) Concerning Present Perfect B for transitive verbs, i.e. the auxiliary have plus a past participle, it is not very common.Most probably, younger generations, including the high school students whose performance is tested in this study, are not likely either to have heard many examples of Present Perfect B with auxiliary have or to have recognized them for what they are.This makes the case of positive transfer from CG an issue that needs a lot of careful thought.Positive transfer, if any, is more likely to have come through their tacit knowledge of SG.This knowledge is not complete, however.To give an example, a large number, if not the majority, of Greek Cypriot university students of Greek language and literature are not aware that in SG past tense cannot have an existential reading and would wrongly use past tense in SG with an existential reading.

Readings associated with the perfect and Universal Grammar
The research presented in this paper was undertaken as part of the general discussion about whether Universal Grammar (UG) is accessed in L2 acquisition.The starting point for the discussion was the assumption that the readings associated with the perfect were root readings, signaled in a number of cases with explicit markers.L2 acquisition of the readings associated with perfect would require establishing connections between these markers and English present perfect.And it is compatible with common sense to assume that this could be how L2 acquisition of the readings associated with the perfect in English takes place.
There is an alternative way one can look at the acquisition (L1 or L2) of the readings associated with the perfect, and this is what we intend to explore in future work.This involves a working hypothesis about how the distinct readings associated with the perfect (across languages) are encoded in UG.As summarized in the Introduction, the three major hypotheses about the role of UG in L2 acquisition differ in that direct access to UG principles is thought to be possible only in the first hypothesis.As far as UG parameters are concerned, direct access is the only possibility in the first hypothesis, one of two possibilities in the second hypothesis, and not at all possible in the third.The obvious question is whether L2 acquisition of the readings associated with the perfect in a language has to do with a UG principle, a UG parameter or it does not have to do with UG at all.If we only take into account the fact that not all languages have a distinction between a present perfect and a past tense, we are likely to conclude that L2, or even L1, acquisition of the perfect has nothing to do with UG.If, however, we make the working hypothesis that the distinct readings associated with the perfect share a core meaning and that, whatever the core meaning of the perfect is, all languages express this meaning somehow, we realize that we should be talking about the basic meaning marked with the perfect, rather than the basic meaning of the perfect.This basic meaning that is taken to be marked with the perfect is presumably some semantic universal.It is not one of the readings associated with the perfect.Rather, it is seen as a semantic universal that triggers/allows the construction of the readings associated with the perfect.Identifying the basic meaning behind all the readings associated with the perfect across languages will also place us in a better position to discuss L2 acquisition of the perfect.And given that in every language it is possible to use either the readings associated with the perfect or readings not associated with the perfect, it is possible to formulate a working hypothesis for the basic meaning marked with the perfect along the following lines.It could, for instance, be that as a semantic universal there are two modes of presenting information and that the perfect, in the languages that make a distinction between a present perfect and a past tense, is the main marker for one of the two modes of presenting information, whereas the other mode is marked with non-perfect verb forms or with some other means.If this is so, the meaning marked with the perfect is more 'sentential' than 'verbal' in its scope.This is for the time being a type of a working hypothesis, and it is beyond the scope of the present paper to make it more concrete.It is our aim, however, in future work to discuss L2 acquisition of English present perfect interpretations by CG speakers in the light of a proposal of this type.
Apart from the basic meaning marked with the perfect, there are the different readings associated with the perfect, in essence compatible with the perfect.The obvious question is how these readings are derived.For that we need to take into account the full range of readings associated with the perfect across languages.It needs to be examined whether the various readings associated with the perfect form a system, and how they do that, or they are disparate interpretations that just happen to co-occur in language after language, which is most unlikely.If the readings associated with the perfect in fact form a system (cf.the proposal in section 2.2.), it could be the case that what is encoded in UG is some semantic universal, i.e. the basic meaning marked with the perfect in terms of the discussion in the previous paragraph, plus the distinct independent readings combined with the perfect, which also involve universal semantic distinctions.In this picture the distinct readings associated with the perfect must result from the interplay between the core meaning marked with the perfect and these independent readings.To sum up, in the outlined research programme we should first identify the semantic universal that constitutes the core meaning that is marked in some languages with the perfect.Then we should identify 'how' this semantic universal triggers the cluster of readings associated with the perfect in some languages.This second part was sketched in section 2.2.For instance, a possible answer could be that the distinct readings are derived through the interaction of the core meaning with aspectual class, but we do not think this is the case, as different aspectual classes are compatible with the same reading (cf. the existential reading).In fact it seems that we will need to posit interaction of the core meaning with two different distinctions.Interaction with the first distinction will yield the readings associated with the perfect in all languages that have both a present perfect and a past tense.Interaction with the second distinction will yield the cross-linguistic differences in the range of readings associated with the perfect.To give a precise example of this, SG present perfect cannot express the "hot news" reading.Consider the English example of the "hot news" reading in (20), the ungrammatical corresponding SG example with a present perfect in (21), and the grammatical equivalent SG example in ( 22).As shown by ungrammatical (21), SG does not associate the hot news reading with the perfect.As shown by grammatical ( 22), the hot news reading is expressed in Standard Greek with a verb-initial sentence in past tense.
Granting that the core meaning marked with the perfect is part of UG amounts to saying that although the perfect is not present in all languages, the readings it is associated with are somehow represented in all languages.However, the readings associated with the perfect as such need not be part of UG.It could be, for instance, that these readings are the result of an interplay between the particular mode that can be marked with the perfect and some other parameter.The fact that languages that distinguish between a present perfect and a past tense do not all have the same range of readings associated with the perfect also suggests that there is an interplay between at least two 'factors' for the derivation of the readings associated with the perfect.
All this takes place in first language acquisition.It can then be investigated whether UG is activated in second language acquisition, or even in second dialect acquisition.The type of working hypothesis sketched above is work in progress.Its relevance here is that if this is the right picture, it would be very difficult for a second language learner who has 'very little' perfect in his native dialect, i.e. the resultative reading of Present Perfect B in CG, to develop an intuition of what is the particular core meaning that is expressed in his dialect mainly by means other than the perfect, and after realizing what that core meaning is make the connection that this core meaning is expressed in English mainly with the perfect, and start using the English perfect in all the appropriate cases.Our feeling is that this could only be possible, if at all, at the later stages of L2 acquisition.It is far more likely that for most part of his L2 learning experience the L2 learner will only be able to identify markers for particular readings of the perfect and learn to associate the perfect verb forms with these particular markers.And that was the basis of the discussion in Section 4.4.
The crucial (for us) working hypothesis that languages that do not have a distinction between a present perfect and a past tense can also somehow express the readings associated with the perfect is compatible with the general assumption in modern linguistics that all languages are equal also in the sense that it is possible to express all meanings in all languages, presumably because the types of meanings are semantic universals.Agouraki (2006) has shown that readings associated with the perfect in English are marked in CG with adverbial modifiers or stress on the verb.In particular, that the existential reading can be marked with adverbials specifying number of times or stress on the verb.In terms of the revised typology for readings associated with the present perfect presented in Section 2.2., it can be said that the indefinite/ characterizing reading with a superimposed quantificational interpretation is marked with adverbials or polarity items denoting number of times (cf.thkio fores 'two times', potte 'ever').The markers in association with the context signal which reading obtains in each case.It should be noted, however, that the same markers are used in English, over and above the perfect, to mark the same readings as in Cypriot Greek.This fact supports, in our view, the claim that the perfect marks the core meaning of all the readings associated with the perfect.
In the case of the perfect we do think that it is possible for UG to be activated, especially in the context of formal teaching, as it is carried out nowadays.Any alleged effort of the L2 learner to acquire the English perfect through UG is undermined by the formal teaching of English.The hypothesis about the possibility of direct access to UG can only be assessed for those who would learn English without formal teaching (for instance, Greek Cypriot immigrants in the U.K. acquiring English).
Finally, it should be pointed out that, although CG lacks Present Perfect A, it has Past Perfect A. The interpretations of Past Perfect A in CG should be investigated, as well as what this knowledge implies for L2 acquisition of English present perfect, and what this fact can tell us about how the readings associated with the perfect are encoded in UG.
distinguish among a universal reading, an experiential/existential reading, a resultative reading and a recent past reading (cf.Iatridou et al.'s respective examples in 1(a)-(d), respectively).(1) a.I have been sick since 1990.b.I have read Principia Mathematica five times.c.I have lost my glasses.d.He has just graduated from college.Portner (2003), on the other hand, distinguishes among a resultative reading, an existential reading, a continuative reading and a hot news reading (cf.Portner's examples in 2(a)-(d), respectively).
differences between the two typologies could be summarized as follows.(a) The sentences where Portner assigns the perfect a resultative reading or a current relevance reading seem to constitute a premise in a logical argument with a tacit conclusion.This 'use' of the readings associated with the perfect is absent from Iatridou et al.'s typology.Both examples 2(a) and 2(b) seem to involve an existential reading with a superimposed 'use' as a premise in a logical argument.This is, in a sense, described by Portner with respect to example 2(b), for which he says that it has both an existential reading and a current relevance reading.So, at least on the basis of the examples used, Iatridou et al.'s resultative reading is used in the literal sense that the result of the event/action described by the predicate continues in the present, while Portner's resultative reading is used in a figurative (i.e.logical argument) sense.(b) Portner's typology lacks the recent past reading while Iatridou et al.'s typology lacks the hot news reading.(c) Portner distinguishes between an existential reading and an experiential reading.Portner's experiential reading appears to correspond to both the resultative reading and the recent past reading in Iatridou et al.'s typology.

( 12 )
In 1989 alone I took my theory test three times.Another reason is that by decomposing the so-called existential reading we can understand where the difference lies among example 10(a), where the definite reading (with a more definite past time) is combined with the quantificational interpretation, example 10(b) and examples 11(a)-(b).

(
14) a.I have been to the States since 2001.b.I was in the States since 2001.

Figure 1 :
Figure 1: Performance of all the pupils in all the tests

( 20 )
Richard Holbrook has died at the age of 69.(21)*o Richard Holbrook ehi pethani se ilikia 69 eton the Richard Holbrook-NOM has-3SG died at age-ACC of 69 years-GEN (22) pethane se ilikia 69 eton o Richard Holbrook died-3SG at age-ACC of 69 years-GEN the Richard Holbrook-NOM "Richard Holbrook has died at the age of 69."

Table 1 :
Performance of all the pupils in each one of the tests

Table 2 :
Performance of all the pupils in all the tests

Table 3 :
Two-sample t-test between the groups

Table 4 :
Test performance by groups Figure 2: Test performance by groupsIf Cypriot private and government schools are considered as one group, there is no statistically significant difference (two-sample t-test) between the two types of schools:

Table 5 :
Two sample t-test between the groups .

Table 6 :
Test production by all groupsFigure 3: Test production by all groups Table 7: Two-sample t-test (primary-gymnasium-lyceum)