Does a second language develop like a tree?
Fractal growth in second language development
Honours Dissertation Final Version Master in English
Faculty of Arts University of Groningen
Margreet van Koert S1332007
Acknowledgments
First of all, I would like to thank my supervisor, Marjolijn Verspoor, to include me in her further analysis of students’ texts written in the OTTO-project. I am grateful that she turned me into an enthusiast about verbs, sentence constructions and Cognitive Grammar. I would also like to thank the people who worked in the OTTO-project, for they made it possible for me to carry out this study.
Finishing this thesis took a greater amount of time than I had anticipated beforehand, but I am grateful for all the experiences – be they linguistics-related or not – that I had during the writing process.
I would like to express my thanks to Arina Banga for having me present my data during an Acquisition Lab in Nijmegen. And thanks to Angeliek van Hout for letting me practice that presentation in the Acquisition Lab in Groningen. I am indebted to my audience for their questions, advice and remarks.
Furthermore, I would like to express a sincere appreciation of my friends, colleagues and fellow students, since they helped shaping this thesis by correcting errors and pointing out sections that needed revision.
From the bottom of my heart I would like to thank my family: my parents, my brothers and sister, their partners and their children.
Explanatory list of terms and abbreviations
A Adjunct
Aux Auxiliary Verb
BO Benefactive Object
Cop Copula Verb
Ditr Ditransitive Verb
DO Direct Object
DST Dynamic Systems Theory
Int Intransitive Verb
IO Indirect Object
L1 First Language
L2 Second Language
L2 learner Second Language Learner
Lex Lexical Verb
Mono Monotransitive Verb
NS Native Speaker
OTTO-project Onderzoek TweeTalig Onderwijs project
(Bilingual Education Research)
P Predicator
S Subject
SA Subject Attribute
SLD Second Language Development
TTR Type Token Ratio
V Verb
Summary
Contents
Acknowledgments 1
Explanatory list of terms and abbreviations 2
Summary 3
Chapter 1 – Introduction 6
Chapter 2 – Background 8
2.1 Usage-Based Grammar 8
2.2 Dynamic Systems Theory 11
2.3 Construction Grammar 14
2.3.1 The Five Basic English Sentence Patterns 15
2.4 Research Questions 17 Chapter 3 – Method 21 3.1 Subjects 21 3.2 Materials 22 3.3 Analyses 22 3.3.1 Syntactic Analyses 23 3.3.2 Construction Analyses 23
3.3.3 Analysis of be and have 24
3.3.4 Analysis of finite verb forms 25
Chapter 4 – Results 27
4.1 Use of constructions and verbs 27
4.1.1 Copula 28
4.1.2 Intransitive 29
4.1.3 Monotransitive 32
4.1.4 Ditransitive 35
4.1.5 Complex-transitive 37
4.2 Use of be and have 38
4.2.1 Be 38
4.2.2 Have 41
5.1 Use of constructions and verbs 49 5.1.1 Copula 50 5.1.2 Intransitive 51 5.1.3 Monotransitive 52 5.1.4 Ditransitive 53 5.1.5 Complex-transitive 53 5.1.6 Constructions 54
5.2 Use of be and have 54
5.2.1 Be 54
5.2.2 Have 55
5.3 Use of tenses 55
5.4 Verb frequencies and TTRs 56
Chapter 1 – Introduction
To equate second language development to the growth of a tree seems a daring step to take. But let us have a look at a tree: do its roots grow first, long before its trunk, its branches or its bark grows? No, all features of a tree grow at the same time. Now let us have a look at second language: does vocabulary grow without developing grammar or without practicing pronunciation? Normal second language acquisition does not, because all these aspects develop simultaneously. The question is whether these aspects develop similarly to one another. Fractals are self-similar shapes that can be found in nature: cauliflower and broccoli consists of tiny florets that are shaped similarly to the whole cauliflower or broccoli. Can these self-similar shapes, or fractals, be found in language development? Do different aspects of language develop similarly? To answer these questions the present study examines second language learners’ texts at seven levels, from 0 (absolute beginners) to 6 (high intermediates). The best thing would be to follow individuals over time, but it would have required a very dense database which is not available at this moment. Instead, we look at 10 individuals’ texts at each level. These texts are analyzed on the following aspects: the sentence patterns, different verbs used in these sentence patterns, different uses of be and have, and the application of tense to determine whether each reveals fractal growth.
Usage-Based Grammar assumes that input frequency affects language acquisition. Therefore, we predict that the most frequent sentence patterns – copula, intransitive and monotransitive – are used by the level 0 to 2 students, whereas level 5 and 6 students will also use the ditransitive and complex-transitive (see section 2.3.1 for a description of the sentence patterns). Furthermore, level 0 to 2 students are believed to make use of a smaller set of verbs than level 5 and 6 students. The verbs used in each construction will show a Zipfian distribution (Zipf, 1935; Larsen-Freeman and Cameron, 2009), which means that there is one verb that accounts for the largest share of tokens in that construction. There is an exponential decrease in the use of other verbs. This distribution is predicted to be found in the development of all aspects: use of constructions, different verbs used in the constructions, different uses of be and have, and uses of different tenses. If this is the case, then there is fractal growth in these subsystems of second language development.
Chapter 2 – Background
Language development is a process of change; some theories describe the factors that contribute to change (e.g. Usage-based grammar), others focus on the process or how language changes (e.g. Dynamic Systems Theory), and still others what changes (e.g. Construction Grammar). As this study examines the factors, the process of change and the constructions that change, these theories are explained below.
2.1 Usage-Based Grammar
In this perspective, language acquisition could be equalled to learning how to behave in a particular manner in a given situation. To this end, children need to replicate specific patterns and structures they hear in the input. Therefore, experiencing language use is necessary for children in order for patterns and constructions to emerge in their language. Language input consists of much repetition and variation, as well as frequent and less frequent words and structures. There are two types of frequencies: token frequency – the number of times a unit appears in a given text or utterance – and type frequency – the number of distinct items which are represented by the pattern (Bybee, 2008). Frequency of items in the input has an enormous effect on language acquisition (Lieven and Tomasello, 2008; Ellis and Ferreira-Junior, 2009), but frequency is not the only predictor of an item being learnt first. Semantic complexity of a structure may impede the speed with which it is learnt. Furthermore, structures which carry similar meanings to other structures that have already been mastered may show a delay in acquisition (Steinkrauss, 2009). Thus, frequency, semantic complexity and a structure’s meaning are all collaborating factors in language acquisition.
As children start out with fully lexical-specific constructions, it takes some time before children abstract away to slot-filling structures. These item-based patterns become increasingly schematic in the child’s representation and eventually the child learns how to express communicative functions in more complex manners (Lieven and Tomasello, 2008). It is not only the schematic representations, but also the combining of pieces of language with new pieces of language that comprises language development. Eventually what changes is the complexity of constructions (how many parts there are), the degree of abstractness in constructions (how far the slots that can be filled with other words reach) and the network of interconnections (Lieven and Tomasello, 2008). In other words, language development begins with fully specific items and local connections and grows towards abstractions and manifold interconnections. Language consists of “constructions” at many levels (morphemes, words, lexemes, sentence patterns and so on) and at all these different levels language will grow.
Input frequency is one of the main factors that influences children’s first production of specific items. This is expected as they can only pick up those words they have heard. Some words appear much more frequently than others, e.g. dog is more frequent than
frequently first. This frequency phenomenon is described in Zipf’s law, which states that in every language only a few words that appear very frequently account for the most linguistic tokens (Zipf, 1935; Ellis and Larsen-Freeman, 2009). Zipf’s law explains that there is an exponential decrease from the most frequent to the second most frequent word. In other words, if the most frequent word appears on average after every 10 words, then the second most frequent word appears on average after every 20 words, and the third after every 30 words, and so on (Zipf, 1935; Larsen-Freeman and Cameron, 2008). Like other power-laws, Zipf’s law is based on large corpora, which means that it is highly predictable, because of the large numbers. Zipf’s law does not only hold for words, but for verb argument constructions, too (Goldberg, 2006; Ellis and Ferreira-Junior, 2009). Some verbs occur very frequently in some constructions, such as give in the ditransitive construction, e.g. Mary gives
John a book. When an exemplar, such as give, is highly frequent in a given construction, it will
contribute to defining the category and it will very likely be considered the prototype (Ellis and Larsen-Freeman, 2009). In this case, give may be prototypical of the notion of transfer in the ditransitive. However, it is not only the frequency of usage which defines prototypicality of a verb construction category, it is also the generic quality of the verb semantics that drives prototypicality (Ellis and Ferreira-Junior, 2009). For example, a verb such as send may occur quite frequently in ditransitive constructions and may also be a good example of a ditransitive construction, yet give fits the meaning of any type of transfer, i.e. someone causing someone to receive something (Ellis and Ferreira-Junior, 2009). Therefore, the prototypicality of give depends on its being both frequent and generic.
L1 than helped (Bybee, 2008). A reasonable option for beginning L2 learners is to rely on a certain prototypical verb for a particular construction. This is exactly what Ellis and Ferreira-Junior (2009) found for L2 learners: there is a Zipf-like distribution of verbs used in a certain construction, i.e. the highest frequency verb accounts for the most tokens in a given construction. However, it is not only frequency of usage which defines construction categories. Some verbs commonly occur in one construction and are therefore very reliable and distinctive as cues for this construction. Other verbs – be for example, which can appear as intransitive and copular verb – are more promiscuous and thus they lack reliability as cues for a given construction (Ellis and Freeman, 2009). Concisely, Ellis and Larsen-Freeman put it as follows:
In natural language the type/token frequency distributions of construction islands, their prototypicality and generality of function in these roles, and their reliability of mappings between these, together, conspire to optimize learning (Ellis and Larsen-Freeman, 2009, p.118).
So, frequency, prototypicality, generality and reliability of a verb are all factors that contribute to first and second language learning of constructions.
2.2 Dynamic Systems Theory
Usage based theory focuses on the factors that affect the acquisition and development of different subsystems of the language system, but it does not explain how each of the subsystems changes over time nor how they may interact with each other over time. A theory that deals with how systems and subsystems may change over time is Dynamic Systems Theory (DST).
subsequently linked to concepts. Therefore, these subsystems collaborate in forming language. Secondly, language change is a result of interaction with external influences, i.e. the environment, and internal reorganization (Verspoor and Behrens, 2010). Even an adult L1 speaker may change his/her language, when s/he moves to a different region where, for instance, prepositions are used differently. This in turn may change his/her own language system in that a particular register is added or broader concepts are connected to certain prepositions. All in all, language is a complex system, because it is interconnected with other subsystems and consists of many subsystems itself, and language is a dynamic system, as those subsystems continually interact and keep on changing the whole system.
DST predicts much variability not only between learners or speakers, but also within learners and speakers, because the system is continually subject to change. In order to detect variability, studies need to be microgenetic; thus, focussed on one individual whose language development is measured at multiple time intervals. Verspoor, Lowie and van Dijk (2008) found variability and continuously changing language system in the texts written by an advanced learner of English. Whenever sentence length increased, the number of academic words she used decreased, and vice versa. They also found peaks (U-shaped behaviour) in some constructions such as word length and the use of longer noun phrases. U-shaped learning and non-linear growth both point to variation and variability in the language system. Verspoor, Lowie and van Dijk concluded that variability is necessary for language development, as it will allow learners to try out multiple strategies. When there is a lot of variability in a very short time span, with peaks and regressions, the language system is changing from one attractor state to another (de Bot, Lowie and Verspoor, 2005; Spoelman and Verspoor 2009). During those critical states, just before the system changes, fractals – i.e. self-similar shapes that repeat themselves – occur at the boundaries, because the frequency distribution around the critical point is fractal. Larsen-Freeman and Cameron explain it thus:
Fractals are yielded by power-laws, which signify scale-free relationships, commonly between two variables (Larsen-Freeman and Cameron, 2008). Zipf’s law is an example of a power-law, as it denotes a relationship between the frequency of occurrence of a word and its frequency rank. Zipf’s law states that there is a small group of words that occur very frequently and there is a large group that appears rarely in comparison. Power-laws occur in complex systems, such as language, and when their constants are repeated, fractals are generated. Commonly, fractals are scale-free, which means that there are similar forms at each size or level.
Fractal-generating processes aid first and second language development, because language shows a similar fractal organisation. For instance, every language exhibits a Zipfian-like distribution of word frequencies and this skewed word distribution helps children and adults to find the most prototypical exemplar of a construction, idiom, et cetera. Fractals are important for complex, dynamic systems, for they provide flexibility to change and adjust to new situations (Larsen-Freeman and Cameron, 2008). For language learners this means that when they come across a different word frequency distribution, they can adapt to this new distribution. The variability and variation caused by an individual learner who is dealing with a newly learnt construction will be masked in the large numbers of a Zipfian distribution. Thelen and Smith (1994) refer to this as the grand sweep view and this is different from the microgenetic view that DST takes.
essential to language development and these are expected to occur in all the different subsystems or constructions of the language. Just like a tree that grows its bark, its trunk, its roots and its leaves all at the same time, so grow the subsystems of a language and they develop simultaneously.
2.3 Construction Grammar
Constructions are linguistic patterns that are stored if they appear sufficiently frequently and if their component parts do not predict some aspect of its form or function (Goldberg, 2006). Constructions are essential to language, since they form the building blocks of utterances. It is with constructions that actual expressions are created. Those linguistic patterns consist of memorized pairings of form and function-constructions that link formal properties to a particular communicative function (Goldberg, 2006). The function part of constructions stores information about the use of entire constructions, such as appropriate situations, register and dialect, for example (Goldberg, 2006). Constructions can also be described as form-meaning-use dynamic patterns of language using (Larsen-Freeman and Cameron, 2008), as they are combinations of form, meaning and use and are liable to change in a speaker’s dynamic language system. This description highlights the usage-based origin of constructions, as it entails that patterns arise through communication, i.e. language use. During communication combinations of constructions are continuously formed and it is only when constructions are in conflict that illicit sentences arise. Since constructions can be combined freely, the creative potential of language is infinite (Goldberg, 2006). However, speakers commonly produce constructions which they have registered earlier (Larsen-Freeman and Cameron, 2008). In Hopper’s words (1998), speakers use sedimented sequences from language and, therefore, the combinations of constructions are not completely novel. These sedimented sequences contribute to fluency, idiomatic expressions and familiarity of native-like language (Pawley and Syder, 1996; Larsen-Freeman and Cameron, 2008). All in all, constructions are patterns whose specifics about form and function are stored.
means that not only formal aspects of constructions need to be learnt, but functional aspects, too. Ellis and Ferreira-Junior define constructions thus:
The constructionist framework holds that learning a language involves the learning of its constructions – the units of the linguistic system, accepted as convention in the speech community and entrenched as grammatical knowledge in the speaker’s mind. Constructions specify the morphological, syntactic, and lexical form of language and the associated semantic, pragmatic, and discourse functions. Constructions form a structured inventory of a speaker’s knowledge. They are useful because of the symbolic functions that they serve. It is their communicative functions that motivate their learning. (Ellis and Ferreira-Junior, 2009, p. 370)
So, Ellis and Ferreira-Junior suggest that constructions are symbols that represent form on the one hand, and function and meaning on the other. According to Goldberg (2006) constructions are centred around the verb, as verbs contribute the lion’s share of a construction’s meaning. For children acquiring the L1 and L2 this means that verbs are good cues to sentence meaning; therefore, the early language patterns are usually concentrated around verbs. Goldberg states: “The construction is at least as good a predictor of overall sentence meaning as the verb, but the verb is more predictive than the construction when it comes to subtle aspects of meaning involving manner or means.” (Goldberg, 2006, p. 106). So, constructions yield a more general meaning, while a verb gives a more specific meaning, which is understandable because a verb carries lexical content. From this it is not surprising that verbs that have similar semantics tend to appear in similar constructions. This tendency aids learners in learning constructions and, thus, a language. Furthermore, constructions that are expressed by a single or few verbs are easier to learn for beginning learners than constructions that are conveyed by manifold various verbs (Goldberg, 2006). Consequently, constructions and verbs are tightly connected in the speakers’ and learners’ language representation.
2.3.1 The Five Basic English Sentence Patterns
basic sentence patterns. As Goldberg (2006) and Larsen-Freeman and Cameron (2008) point out, traditional grammatical structures, such as basic sentence patterns are also constructions with representations. The present study will therefore examine the five basic English sentence patterns.
According to Verspoor and Sauter (2000), the five basic sentence patterns in English consist of copula, intransitive, monotransitive, ditransitive and complex-transitive constructions.
Copular verbs are verbs that link an adjective or a noun to the subject of a sentence. Copular verbs do not convey much meaning; they generally express that A is equal to B. The verb frame pattern consists of: Subject (S) + Predicator (P) + Subject Attribute (SA). Common copular verbs are: be, appear, seem, look, sound, smell, taste, feel, become and get. Two examples of a copular verb frame are given in (2) and (3).
(2) Mary feels happy
S P SA
(3) Mary becomes a student
S P SA
Intransitive verbs are verbs that denote an action in which only one actor is present. Many intransitive verbs express a pure action, such as run, swim, cycle and sit. Adverbials may appear in sentences that comprise an intransitive verb, but they are not necessary supplements. The verb frame pattern is as follows: Subject + Predicate (+ Adjunct). An example of an intransitive verb frame is given in (4).
(4) Mary cycled (to London)
S P A
(5) Mary built the castle
S P DO
Ditransitive verbs express events that usually involve three actors: the subject, who gives something to someone or does something for someone, the direct object, which is given or done, and the indirect or benefactive object, who receives something. Common ditransitive verbs are buy, give, lend, pay, read, send, teach and wish. The verb frame consists of the following pattern: Subject + Predicator + Indirect Object / Benefactive Object + Direct Object. Two examples of a ditransitive verb frame are given in (6) and (7).
(6) Mary sent John a letter
S P IO DO
(7) Mary wished John many happy returns
S P BO DO
Complex-transitive verbs are verbs that cause the direct object to belong to a different category. Only a few verbs can occur in this verb frame, such as make, call, crown,
consider, assume and deem. The verb frame is as follows: Subject + Predicator + Direct Object
+ Object Attribute. Two examples of a complex-transitive verb frame are given in (8) and (9).
(8) They crowned Mary queen
S P DO OA
(9) Mary considered the letter sent
S P DO OA
Copula, intransitive, monotransitive, ditransitive and complex-transitive are the five basic patterns, the first three of which are commonly used in everyday language, and the last two appear less frequently (Verpsoor and Sauter, 2000). There are variations on these patterns (see, for example, Swan, 1980; Mackenzie, 1997), but these are outside the scope of the current analysis.
Usage-based grammar explains how experience with language and experience with social situations in which that particular language is used eventually leads to change in the representations of language. Especially when items or structures are frequently used in a language, these become entrenched more quickly than rarely used items or structures. The phenomenon that some items are much more frequently used than others is described in the Zipfian distribution of language (Zipf, 1935; Ellis and Larsen-Freeman, 2009). Zipf’s law is a power-law that yields fractals which are generated just before a complex, dynamic system changes. DST assumes that fractals enable learners to adapt to new frequency distributions. Frequency distributions are also present in the use of sentence patterns or constructions (Ellis and Ferreira-Junior, 2009). Constructions are linguistic patterns that link form to function and meaning (Goldberg, 2006). All in all, the use of constructions changes because of input frequency and fractals which allow a complex, dynamic system such as language to reorganise itself.
The current study focuses on the development of sentence constructions and lexical variety of verbs by Dutch second language learners of English in a cross-sectional analysis. In particular, the five basic English sentence patterns that are used by the L2 learners and the verbs that appear in these patterns are examined. Language development might be captured in the metaphor of a tree that grows, since the many features of a tree grow simultaneously. For instance, as a tree’s trunk is getting taller, so is its bark changing colour, twigs are evolving and its roots are settling deeper into the ground. This can be seen in language development as well, for many aspects – such as vocabulary, tense system, meaning of words – develop simultaneously. However, just like the tree whose trunk gets stronger and thicker, the linguistic structures or words that were learnt and produced first become more and more entrenched and become automatized (Bybee, 2008). Thus the research question is:
Does a second language develop like a tree? In other words, is there simultaneous fractal growth in sentence constructions and the lexical verb variety in those constructions?
(1) Beginning second language learners will make use of few constructions – only the copula, intransitive and monotransitive will occur – as the learners become more advanced they will use the ditransitive and the complex-transitive constructions, too; however, the copula, intransitive and monotransitive will be used more frequently than the ditransitive and complex-transitive.
It is expected that L2 learners start out with the most basic and most frequent constructions, so they will use the copula, intransitive and monotransitive first. When L2 learners have had more experience and more practice in the L2, they advance to include more peripheral constructions, such as the ditransitive and the complex-transitive.
(2) Beginning second language learners will use set patterns in constructions, so there is little variation in the use of verbs. As learners become more advanced, they will expand their verb variety. Yet, the few verbs learners used in the beginning for a particular construction will remain the lion’s share of the total amount of verbs that more advanced learners use.
L2 learners are expected to use the most prototypical and generic verbs in constructions first, as was found by Ellis and Ferreira-Junior (2009). It is probable that learners will use (an inflected form of) be for the copula, go for the intransitive, have for the monotransitive, give for the ditransitive, and call for the complex-transitive. As learners pick up more and more language, they will expand their verb knowledge and show this by employing a greater variety of verbs in these constructions, instead of holding on to the most prototypical and generic verbs. However, it is also expected that those most prototypical and generic verbs will remain to occur frequently in the advanced learners’ texts.
(3) From the lower levels to the higher levels there will be a Zipfian-like growth in use of constructions, variety of verbs in those constructions, in use of tenses, and in use of be and have. However, the individual learner’s growth patterns shows non-linear development, variation and peaks and regressions.
present study are not based on linear progressions. So, a tendency towards the Zipfian-like growth presumed, but a completely linear progression from one level to another is not expected. DST describes that individual development will be non-linear, variable, with peaks and regressions and showing U-shaped development. So, there will be individual patterns that are whimsical, because learners tend to focus on one language aspect at the expense of other aspects. All in all, a continuous growth is expected in all aspects of the L2 from one level to another, which is illustrated in Figure 1.
Figure 1. A simplified model representing the expansion of L2 learners’ use of constructions, verbs or tenses.
Figure 1 is an abstract model that shows the language development of L2 learners. In the following chapters this linear development will be described and explained, although individual L2 development may well show bumpy and whimsical growth and U-shaped development of constructions, verbs and/or tenses.
Advanced L2ers
Chapter 3 – Method
In order to detect whether there is fractal growth in L2 development, several analyses were carried out in the current study and their particulars are described here. The subjects’ details are given, the materials are described and the syntactic analyses, the analyses of the constructions, the be and have analyses and the finite verb form analyses are presented.
3.1 Subjects
In the Onderzoek TweeTalig Onderwijs (OTTO) (Bilingual Education Research) project 481 monolingual Dutch secondary school students from six different schools were asked to write a 200-word-text in their L2, i.e. English, on either their feelings towards their new school – for first-year-students – or their experiences during the summer holidays – for third-year-students (van Rein, 2010). To obtain texts in a wide range of proficiency levels, students from different types of schools participated in the study. Three of the secondary schools offered bilingual education, which means that half of the students’ subjects were taught in English and the other half in Dutch. The other three schools taught all subjects in Dutch. The differences between these schools are not investigated in this study.1
In the OTTO project the students’ texts were first holistically scored on general proficiency level by a team of 8 raters. The texts were categorised into 7 levels: from 0, being the lowest level, to 6, which is the highest level. Descriptions of the levels can be found in Appendix II. These texts were then analysed by means of CHAT and CLAN on the basis of complexity (morphology, syntax and number of clauses), accuracy (grammatical and spelling errors) and lexical variables (authentic chunks, lexical frequency profile and TTR). The complexity, accuracy and lexical variables were qualitatively checked and it was found that there were significant differences between all levels on these variables. So, we can assume that the holistic scores given were quite adequate in assigning levels of language proficiency. Hence, it is more conducive to base the analyses on the different proficiency levels than on different groups (e.g. first-year-students versus third-year-students), because the groups contain many different levels.
For the current study, 70 texts were chosen at random, 10 to represent each level. Ten texts per level were chosen to allow detailed, qualitative analysis per level. In a study
based on DST, a longitudinal study of individual learners would be preferred to a cross-sectional study, but the type of dense data needed for such a study is difficult to obtain. Therefore, longitudinal development is simulated to a degree by taking 10 texts per level. Since the lower levels contained more pieces of writing by first-year-students than the higher levels, a skewed relationship is visible in the sample. In other words, level 0 comprises no third-year-students, whereas level 6 includes no first-year-students. All other levels contain both first- and third-year-students.
Table I.1 in Appendix I lists the student numbers, the year the students were in, their level, the number of students at that level, the number of words the students used, the number of clauses they constructed, and the Type Token Ratio (TTR) of the words they used.
3.2 Materials
During their English classes the students were asked to write a text in English containing about 200 words in an online computer programme which was especially created for the project. The texts were edited to standard CHAT conventions from CHILDES (MacWhinney, 2000) in the OTTO project. Afterwards, the texts were analysed for errors, authenticity, syntax and morphology in the CLAN programme, which is available from CHILDES.
In the present study the students’ raw texts – so not the texts that were edited to CHAT conventions – were used to conduct the analyses. The raw texts were chosen rather than the edited texts, because the previous analyses were not essential to carry out the present analyses.
After the syntactic analyses were conducted, calculations on constructions, verbs, be and have, and tenses were done in Excel.
3.3 Analyses
3.3.1 Syntactic analyses
The syntactic analyses were based on standard English sentence analysis (Verspoor and Sauter, 2000). Sentences from 70 texts were syntactically analysed at clause level (Subject, Predicate, Direct Object, Indirect Object, Subject Attribute, Adjunct) and at word level (pronoun, noun, adjective, adverb, etc.). An example of the syntactic analysis is given in (1).
(1) The first day at school I was very nervous because
A S P|cop be SA A
I didn’t know anyone.
S P|aux do [+ adv] + lex know DO (level 5)
All 70 texts were manually analysed and these data were then entered into Excel. The number of clauses written by each student were calculated, see Table I.1 in Appendix I. The number of different verbs used by each student and for each level were counted. In addition, the auxiliaries, modals and tenses were labelled.
3.3.2 Construction analyses
English has five constructions or sentence patterns and the verb frames from the students’ texts were sorted according to these five categories: copula, intransitive, monotransitive, ditransitive and complex-transitive (Verspoor and Sauter, 2000). The particulars of these constructions are given in section 2.3.1.
In the students’ texts passives which only contained Subjects and Adjuncts were analysed as intransitives, such as I got bitten and I was called by my cousin. Passives which contained Subject Attributes were analysed as copulas, such as one is called Leo. If students used a verb in a non-target verb frame, e.g. she’s learning us good English, the sentence construction was counted, rather than the verb frame designated by the verb. Thus, in this case the sentence was labelled a ditransitive. This manner was chosen, because the present study investigates the use of constructions, the development thereof and the use of verbs in those constructions.
3.3.3 Analysis of be and have
Since be and have occur frequently and carry multiple functions, an additional analysis was conducted on these two prominent verbs. The verb be can appear as copular verb, intransitive verb and as auxiliary, as can be seen in examples (10) – (14).
(10) Mary is happy S P|cop be SA (11) Mary is a student S P|cop be SA (12) Mary is at school S P|lex be A
(13) Mary is cycling to London S P|aux be + lex cycle A
(14) Mary is punched
S P|aux be + lex punch
For the auxiliary it holds that all uses of be, whether in a progressive (13) or in a passive form (14), were labelled auxiliary-be. No examples of be as aux of mood (e.g. he is to
leave soon) occurred in the data.
The verb have can occur as monotransitive, modal and auxiliary verb, which is shown in examples (15) – (17).
(15) Mary has a sister
S P|lex have DO
(16) Mary has to do some homework S P|mod have to + lex do DO
(17) Mary has been to the Bahamas S P|aux have + lex be A
3.3.4 Analysis of finite verb forms
The students’ use of finite verb forms was analysed in a coarse-grained analysis, i.e. only Present, Past, Perfect and Modal were distinguished. In addition, an error analysis was carried out.
Present included simple present active voice, simple present passive voice, simple present dummy verb do with an infinitive, present progressive active voice and present progressive passive voice. Examples are given in (18a) – (18e).
(18) a. Mary writes b. A letter is written c. Mary does not write d. Mary is writing
e. A letter is being written
Past consisted of simple past active voice, simple past passive voice, simple past dummy verb did with an infinitive, past progressive active voice and past progressive passive voice. Examples can be seen in (19a) – (19e).
(19) a. Mary wrote b. A letter was written c. Mary did not write d. Mary was writing
e. A letter was being written
Perfective included present perfective active voice, present perfective passive voice, past perfective active voice and past perfective passive voice. Examples are given in (20a) – (20e).
Modal comprised modal present active voice, modal present passive voice, modal past active voice and modal past passive voice. Examples are shown in (21a) – (21e).
(21) a. Mary can write b. A letter can be written c. Mary could write d. A letter could be written
Chapter 4 – Results
The use of constructions and the number of different verbs were analysed. In addition, an analysis was done on the expansion of the verbs be and have, since they can appear as auxiliary, copula and intransitive and as auxiliary, modal and monotransitive, respectively. Finally, the tenses of the finite verbs were analysed.
4.1 Use of constructions and verbs
There were five types of constructions found in the students’ texts: copula, intransitive, monotransitive, ditransitive, and complex-transitive. In these constructions different verbs were used. The number of these different verbs did not only differ per level – which is shown in the next sections – but also per type of construction, as some constructions allowed more different verbs in the predicate verb frame than others.
0% 20% 40% 60% 80% 100% Leve l 0 (n =74) Leve l 1 (n =142 ) Leve l 2 (n =139 ) Leve l 3 (n =130 ) Leve l 4 (n =155 ) Leve l 5 (n =219 ) Leve l 6 (n =248 ) complex-transitive ditransitive monotransitive intransitive copula
4.1.1 Copula
The first type of construction which was analysed was the copula construction. As can be seen in Figure 1, the copula construction was used somewhat more frequently at the lower levels than at the higher levels.
The percentages of the different verbs used in copula constructions is shown in Figure 2a. 0% 20% 40% 60% 80% 100% level 0 (n=30) level 1 (n=40) level 2 (n=55) level 3 (n=39) level 4 (n=47) level 5 (n=50) level 6 (n=63) copula Turn Seem Go Fall Stay Feel Be like Get Call Look Like Become Be
Figure 2a. The types of verbs in the copula construction.
For level 0 there were nine students who used 30 copula constructions in total and be was used in all copula constructions. At level 1 all ten students used copula constructions and at all times with be. Then at level 2 all ten students used copula constructions and, apart from one occurrence of go and one of call, be was used at all times. For level 3 there were again ten students who used be at all times in their copula constructions. At level 4 all ten students use
be and there were two occurrences of stay and one of look like and call in this construction.
Together the students used 13 different lexical verbs in the copula construction. These 13 different verbs were not used at each level. At the lower levels only one out of the 13 different verbs was used (i.e. be). To visualize how the number of different verbs out of the total number of different verbs branch out from level to level, Figure 2b shows the percentages of different verbs used per level out of all different verbs used in total.
0% 20% 40% 60% 80% 100%
be be be, go, call be be, stay, look like, call be, become, stay together, feel, get, call be, feel, become, get, be like, seem, look like, turn, fall level 0 level 1 level 2 level 3 level 4 level 5 level 6
Copula
Figure 2b. The percentages of all verbs used per level in the copula construction (n=13).
In total 13 different verbs were used by all students for the copula construction. Until level 3 students mostly used be, whereas from level 4 onwards students began using different verbs in this construction.
4.1.2 Intransitive
The intransitive construction was analysed for all levels. As can be seen in Figure 1, the intransitive construction was used in one fifth of the clauses in levels 0 to 2 and in almost two fifth of the sentence clauses in levels 3 to 6.
0% 20% 40% 60% 80% 100% level 0 (n=14) level 1 (n=28) level 2 (n=30) level 3 (n=52) level 4 (n=54) level 5 (n=87) level 6 (n=94) intransitive other Vs run swim walk sit live get come happen be go
Figure 3a. The types of verbs in the intransitive construction.
For level 0 there were eight students who used intransitive constructions and they wrote 14 intransitive constructions in total with nine different verbs. Most verbs, such as go, live, sit and walk, were used by two students. Be was used by only one student in an intransitive construction. At level 1 seven students used 28 intransitive constructions and they used ten different verbs. Students mostly used go, be, live and come. At level 2 there were nine students who used 30 intransitive constructions together and be and go were used by most students. For level 3 all ten students together used 52 intransitive verb frames and many students used
go, be and swim. At level 4 all ten students together used 54 verb frames in the intransitive
construction. All students used go (out/back/away) and six students used be. Then at level 5 there were 87 intransitive constructions for all ten students and nine students used go, whereas five students used be. Finally, for level 6 all ten students together used 94 intransitive constructions. Most students used go, be, happen, come and get. Overall, the relative numbers of the verbs change: at the lower levels go and be comprise a larger share than at the higher levels.
0% 20% 40% 60% 80% 100%
go, be, live, sit, walk,
etc
go, be, come, live,
etc
go, be, get, etc go, be, happen, come, sit, walk, swim, etc go, be, happen, come, get, swim, run, etc go, be, happen, come, get, live, walk, sit, run, etc
go, be, happen, come, get, sit, walk, swim, run, etc level 0 level 1 level 2 level 3 level 4 level 5 level 6
Intransitive
Figure 3b. The percentages of all verbs used per level in the intransitive construction (n=106).
Table 1. Specifics of Intransitive verbs.2
Intransitive verbs # TTR of lex Vs
Level 0 sit, go, be, live, lie, walk, bike, lunch, speak 9 .64 Level 1 be, go (away), live, ride, come, listen, sit, start, wake up 9 .32 Level 2 be, go, play, do, fall asleep, get, rain, sleep, smile, work 10 .33 Level 3 be, go (down), swim, happen, sleep, amputate, bite, burn, care,
change, come back, die, fall, horseback ride, land, play, ride, sit, talk, tell, trip over, walk
22 .42
Level 4 be, go (down), happen, bite, come back, rain, sail (around), amputate, complain, die, escape, fall down, fly, get, kill, leave, relax, rest, revalidate, rot, run, suffocate, survive, swim, wait, worry
26 .48
Level 5 go (back/out), be, come back, happen, drive, live, walk, kiss, say goodbye, sit, wake up, call, chat, cry, die, explode, fall asleep, freak out, get, jump out, look, puke, relax, run, sleep, stay, steal, take, turn, worry
30 .34
Level 6 go (back), be, happen, get (out), come (out/back), dive, walk, arrive, hurt, jump off, lie, run (back), breathe, call, care, disappear, dive, drop, fall, go wrong, know, look, poop out, rush, say, scrape, see, shine, sit, stand up, start off, stream, stutter, sunbathe, swallow, swim, tell, trip, turn around, wait, wake up, warm up, wink
43 .46
4.1.3 Monotransitive
For all levels, the monotransitive construction was analysed. As can be seen in Figure 1, the intransitive construction was used in more than one third of the clauses by students from all levels.
The verbs used in this construction are shown in Figure 4a.
0% 20% 40% 60% 80% 100% level 0 (n=29) level 1 (n=71) level 2 (n=53) level 3 (n=36) level 4 (n=52) level 5 (n=76) level 6 (n=83) monotransitive other Vs love make say think do get see know like have
Figure 4a. The types of verbs in the monotransitive construction.
For level 0 there were nine students who in total used 29 monotransitive constructions with eight different verbs. Eight students used have (got); other verbs used at this level were like, do,
say, love, wait (a substitute for know from the Dutch weten ‘to know’), hate and make. At level 1
there were nine students who together used 71 monotransitive constructions comprising 20 different verbs. The most frequently used verbs were have, like and know. At level 2 eight students used 53 monotransitive constructions with 21 different verbs. Most students used
have, like, tell, know and make. For level 3 there were eight students who used 36
Not only did the absolute numbers of monotransitive constructions for the levels increase, but the number of different verbs increased too. Moreover, the relative numbers changed, too: at the lower levels have is clearly the most frequent verb, but at the higher levels have is not such a prominent verb in the monotransitive construction anymore. Furthermore, students at levels 5 and 6 used verbs that are stereotypically used in ditransitive or intransitive constructions in a creative way in monotransitive constructions. For example, ‘“yeah”, she laughed’ (level 6) for an intransitive verb in a monotransitive construction and ‘I took her hand’ (level 5) for a ditransitive verb in a monotransitive construction.
Figure 4b shows the differences between percentages of the variety of verbs used per level. 0% 20% 40% 60% 80% 100% have, like, do, say, love, etc have, like, know, see, get, do, think, say, make, love, tell, etc have, like, know, see, get, do, think, say, make, take, love, tell, meet, hear, want, etc have, like, know, see, get, do, think, make, tell, want, etc have, know, see, get, do, think, make, take, love, meet, etc have, like, know, see, get, do, think, say, make, take, love, decide, meet, hear, want, etc have, like, know, see, get, do, think, say, make, take, decide, hear, want, etc level 0 level 1 level 2 level 3 level 4 level 5 level 6
Monotransitive
Figure 4b. The percentages of all verbs used per level in the monotransitive construction.
very diverse and usually appear only once, such as take care of, manage, guide, surround and spend. Table 2 gives the specific verbs and the number of verbs used at each level.
Table 2. Specifics of monotransitive verbs.
Monotransitive verbs # TTR of lex Vs
Level 0 have (got), like, love, do, wait, say, hate, make 8 .28 Level 1 have (got), like, do, know, love, get, make, see, learn, say, bring,
can, find, mean, speak, tell, think, trust, understand, write
20 .28
Level 2 have (got), like, hate, know, see, tell (about), think, get, make, talk with, do, forget, hear, hope, love, meet, miss, say, take, throw, want
21 .40
Level 3 have, like, get, see, think, buy, begin with, bite, do, hope, know, learn, make, tell, want, wear
16 .44
Level 4 have (got), do, know, meet, see, speak, think (of), like, ask for, be worth, buy, decide, enjoy, forget, get, give, have in mind, hit, hope, kill, learn, lose, make, play, take, throw, want
27 .52
Level 5 see, have, get (back), like, find (out), say, think, hear (about), know, shoot, take, decide, meet, ask, believe, catch, check for, do, eat, end up with, enjoy, forget, hate, hope, level, love, make, manage, miss, plan, survive, switch on, take care, throw, want
35 .46
Level 6 say, do, have (on), get (on), think, want, discover, find (out), know, like, try, decide, eat, hide, lash, laugh, see, start, surprise, take, allow, ask, bring, enjoy, follow, forget, guide, hear, kiss, learn, make, miss, open, reach, scream, speak, spend, surround, understand, use
40 .48
4.1.4 Ditransitive
0% 20% 40% 60% 80% 100% level 0 (n=0) level 1 (n=2) level 2 (n=0) level 3 (n=3) level 4 (n=2) level 5 (n=3) level 6 (n=8) ditransitive Spare Sell Learn Ask Give Tell (about)
Figure 5a. The types of verbs in the ditransitive construction.
For levels 0 and 2 there were no students who made use of the ditransitive construction. In level 1 there were two students who both used a ditransitive construction once. The two verbs used at level 1 were learn and give. The former cannot be used in a ditransitive construction in English, but its translation equivalent can in Dutch (leren). The latter, give, and its Dutch translation equivalent can occur in ditransitive constructions. At level 3 there were two students who made use of three ditransitive construction. Both verbs, give and tell were used correctly in this construction. Interestingly, the Dutch translation equivalent of tell ‘vertellen’ is also typically used in ditransitive constructions. For level 4 there was one student who used both tell and sell in two ditransitive constructions. Again, the Dutch translation equivalent of sell ‘verkopen’ can occur in a ditransitive verb frame. At level 5 there were two students who made use of three ditransitive constructions and they used tell. Finally, for level 6 there were four students who used eight ditransitive constructions including the verbs tell,
spare and ask.
0% 20% 40% 60% 80% 100%
learn, give give, tell about
sell, tell tell tell, spare, ask level 0 level 1 level 2 level 3 level 4 level 5 level 6
Ditransitive
Figure 5b. The percentages of all verbs used per level in the ditransitive construction.
In total six different verbs were used in the ditransitive construction. Students started out with familiar verbs from Dutch, such as learn and give and then branched out to tell, sell, spare and ask. The ditransitive construction is not a very frequent construction in English (Verspoor and Sauter, 2000) and only few verbs can appear in this construction, hence it is expected that second language learners start to incorporate this particular construction at later stages in their acquisition path.
4.1.5 Complex-transitive
For all levels the complex-transitive constructions were analysed. As can be seen in Figure 1 very few complex-transitive constructions are used. Therefore, instead of using a graph the use of complex-transitive constructions is merely described here.
At level 0 one student wrote ‘I found it well funny’, which may seem authentic British English, but could represent a literal translation from Dutch ik vond het wel leuk ‘I
thought it was rather OK’. A similar construction based on Dutch was used at level 1 ‘we don’t
appreciate that’. Again, at level 2 the only complex-transitive construction found in the
students’ texts was based on Dutch ‘that make’s [sic] my bag very lift [light] like a feather’, which translates to Dutch as dat maakt m’n tas erg licht als een veertje ‘it makes my bag light as a
feather’. So, these complex-transitive constructions were probably based on Dutch. At the
higher levels the complex-transitive constructions seemed to be based more on the L2. At level 5 there were three complex-transitive constructions, one of which was used for ‘history I all so [sic] find interesting’. Another student wrote: ‘Some people might call me a nerd, others may call me a god’. Finally, at level 6 one student wrote ‘we discovered something that made us a little scared’ and another wrote ‘which made us miss the flight!’. Overall, the complex-transitive construction is used seldom and only with three verbs: find, make and call.
Since there were so few verbs used in this construction, no figure is shown which represents the percentages of all verbs used per level.
4.2 Use of be and have
Both be and have can occur in three different constructions. The verb be can appear in auxiliary, copula and intransitive constructions and have can appear in auxiliary, modal and monotransitive constructions. This section explores the change in use of these constructions for be and have over the levels.
4.2.1 Be
0% 20% 40% 60% 80% 100% level 0 (n=31) level 1 (n=55) level 2 (n=70) level 3 (n=64) level 4 (n=59) level 5 (n=73) level 6 (n=82) BE Intransitive Auxiliary Copula
Figure 6a. The uses of be per level.
At level 0 there were ten students who used be, nine of them used be in 30 copula constructions and one of those nine also used be in intransitive constructions. For level 1 all ten students used be, all of them in copula constructions, three in auxiliary + lexical verb constructions and three other students in intransitive constructions. Seventy occurrences of
be were found at level 2, ten students made use of be in copula constructions, four of them in
Overall, it is clear that copula-be yields the most prominent usage of be; the use of intransitive-be increases from level 0 to 3 and from then on its use decreases somewhat. Finally, occurrences of auxiliary-be increase at the higher levels, too.
Figure 6b shows the percentages of the constructions used overall per level.
0% 10% 20% 30% 40% 50% level 0 (n=31) level 1 (n=55) level 2 (n=70) level 3 (n=64) level 4 (n=59) level 5 (n=73) level 6 (n=82) BE Cop (n=300) Aux (n=54) Int (n=81)
Figure 6b. The percentages of constructions used per level for be.
4.2.2 Have
Of all 70 students there were 56 who used have in the auxiliary, modal and/or monotransitive construction. Figure 7a represents the relative numbers of have-constructions per level.
0% 20% 40% 60% 80% 100% level 0 (n=20) level 1 (n=23) level 2 (n=17) level 3 (n=17) level 4 (n=17) level 5 (n=21) level 6 (n=19) HAVE Modal Aux Mono
Figure 7a. The uses of have per level.
At levels 0 to 2 have mainly occurred in monotransitive constructions. From level 3 onwards the auxiliary use and the modal use of have increase considerably. At level 0 there were eight students who used have in monotransitive constructions and another student used have to as a modal.3 For level 1 there were eight students who made use of have in monotransitive constructions, one of them also used have as an auxiliary and another student used have merely as a modal. At level 2 there were six students who used have in monotransitive constructions and two students made use of have as an auxiliary. At level 3 there were four students who used have for the monotransitive construction, four students who made use of
have as a modal and three students who used have as an auxiliary. For level 4 there were four
students who made use of have in monotransitive constructions, six students used have as an
auxiliary and three used have as a modal. At level 5 there were five students who put have in monotransitive constructions, five who used have as an auxiliary and four who made use of
have as a modal. Finally, at level 6 four students used have in monotransitive constructions,
five used have as an auxiliary, and five used have as a modal.
Figure 7b shows the percentages of the overall used constructions per level.
0% 10% 20% 30% 40% 50% level 0 (n=20) level 1 (n=23) level 2 (n=17) level 3 (n=17) level 4 (n=17) level 5 (n=21) level 6 (n=19) HAVE Mono (n=84) Aux (n=28) Mod (n=22)
Figure 7b. The percentages of constructions used per level for have.
The graph in Figure 7b is from 0% to 50% and shows the use of monotransitive-have, auxiliary-have and modal-have per level divided by the overall number of monotransitive-haves, auxiliary-haves and modal-haves. In total 84 monotransitive-haves were used and 19% to 25% occurred at levels 0 to 2, whereas 6% to 10% appeared at levels 3 to 6. All students used 28 auxiliary-haves, 12% of which were used at levels 0, 1 and 2, but at levels 3 to 6 the percentages fluctuate between 17% and 29%. Modal-have was used 22 times, 5% at levels 0 and 1, whereas at levels 3, 4, 5 and 6 the percentages increased and oscillated between 9% and 40%.
4.3 Use of tenses
All finite clauses were analysed with regard to the tenses of the finite verbs and concerning correct use of tense and correct spelling. The exhaustive fine-grained analysis of all tenses is given in Table IV.1 in Appendix IV. Figure 8a shows the coarse-grained analysis in which, for example, all present tenses (simple present and present progressive in active and passive voice) are combined for present.
0% 20% 40% 60% 80% 100% leve l 0 (n =74) leve l 1 (n =142 ) leve l 2 (n =139 ) leve l 3 (n =130 ) leve l 4 (n =155 ) leve l 5 (n =219 ) leve l 6 (n =248 ) Mod Perf Past Pres
Figure 8a. The percentages of the use of tenses per level.
Overall, present tense usage decreased, whereas past tense usage increased. Furthermore, there seems to a sharp distinction between levels 2 and 3 with regard to present and past tense usage. The analysis of tenses per level is as follows.
At level 0 there were 82 predicates in total and 74 of these included English tense, the remaining verbs were Dutch. The greater part, namely 68, of these verbs carried present tense inflection, whereas only three had past tense inflection. There was one present perfect and two modal constructions.
At level 2 there were 139 predicates in total. Of these predicates 107 verbs appeared in the present tense and 20 verbs in the past tense. There were two verbs with present perfect tense and ten verbs in a modal construction.
For level 3 there were 130 predicates in total. There appeared 40 verbs in the present tense and 77 verbs in the past tense. Perfect constructions were used 8 times and modal constructions 5 times.
At level 4 there were 155 predicates. There were 33 verbs that appeared in the present tense and 103 verbs in the past tense. Perfect constructions occurred five times and there were fourteen modal constructions.
For level 5 there were 219 predicates. The present tenses were used for 32 verbs and the past tenses for 156 verbs. There five perfect constructions and 25 modal constructions.
At level 6 there were 248 predicates. Of those predicates 23 verbs appeared in the present tense, whereas 196 carried past tense inflection. Perfect constructions occurred eight times and modal constructions 21 times.
Mistakes in use of tense or spelling errors occurred, too, and Figure 8b shows the students’ errors. 0% 5% 10% 15% 20% 25% leve l 0 (n =74) leve l 1 (n =142 ) leve l 2 (n =139 ) leve l 3 (n =130 ) leve l 4 (n =155 ) leve l 5 (n =219 ) leve l 6 (n =248 ) *Mod *Perf *Past *Pres
Figure 8b is from 0% to 25% because the share in mistakes and errors did not surpass 25% for any level. As there are more present tenses at the lower levels, there are also more mistakes in the present tense at the lower levels than at the higher levels. On the other hand, incorrect usages overall drop over the levels. So, there were relatively more mistakes in tense and spelling at levels 0, 1 and 2 than at levels 5 and 6.
4.4 Verb frequencies and TTRs
Table 3. Frequency details of copula, intransitive and monotransitive constructions.
TTR copula TTR intransitive TTR monotransitive % present tense
Level 0 .03 (1/30) .64 (9/14) .28 (8/29) 92% 3 most frequent verbs be (30) sit (4) go (2) live (2) have (18) like (3) love (3) Level 1 .04 (1/40) .32 (9/28) .28 (20/71) 86% 3 most frequent verbs be (40) be (9) go (8) live (4) have (21) like (15) do (4) Level 2 .05 (3/55) .33 (10/30) .40 (21/53) 76% 3 most frequent verbs be (53) call (1) go (1) be (12) go (9) play (2) have (14) like (10) hate (3) Level 3 .03 (1/39) .42 (22/52) .44 (16/36) 30% 3 most frequent verbs be (39) be (16) go (8) swim (5) have (8) like (5) get (4) Level 4 .09 (4/47) .48 (26/54) .52 (27/52) 21% 3 most frequent verbs be (43) look like (2) stay (1) be (12) go (11) happen (4) have (9) do (5) know (5) Level 5 .12 (6/50) .34 (30/87) .46 (35/76) 14.5% 3 most frequent verbs be (44) become (2) feel (1) go (23) be (17) come back (7) see (9) have (7) get (5) Level 6 .14 (9/63) .46 (43/94) .48 (40/83) 8.5% 3 most frequent verbs be (51) be like (2) become (2) go (18) be (14) happen (7) say (8) do (5) have (5)
construction, but for levels 1 to 4 it is be which appears to be prototypical and for levels 5 and 6 it is go.
There is more of a pattern in the monotransitive construction: as proficiency increases, the TTRs of the levels slowly increase. Furthermore, have appears to be the prototypical verb for levels 0 to 4, whereas for levels 5 and 6 there seem to be many verbs which can be used in this slot.
The TTRs of the copula constructions remain very low for every level, but they do increase over the levels. So, the most prototypical verb is be for all levels, but from level 4 onwards other verbs are used, too, in this construction.
Chapter 5 – Discussion
The current study investigated the sentence construction development and lexical variety of verbs in second language learners’ texts. The learners were native speakers of Dutch who were in their first or third year of secondary school and their second language was English. The sample consisted of 70 texts written in English by the learners. The learners were divided into seven proficiency levels and ten texts per level were analysed. The sample of texts is cross-sectional, but the levels are assumed to simulate to a degree L2 developmental stages. The starting point in analysing the learners’ sentence constructions were the five basic English sentence patterns: copula, intransitive, monotransitive, ditransitive and complex-transitive. In addition, the lexical variety of the verbs in these constructions per level was investigated. Moreover, the multiple functional usages of be and have were examined together with the usage of tenses to determine whether besides constructions and verbs (shown by e.g. Goldberg, 2006; Ellis and Ferreira-Junior, 2009) other aspects of an L2 also show a Zipfian-like growth during development.
Three hypotheses regarding the learners’ development in these texts were formulated. The hypotheses are stated in section 2.4 and they are repeated here.
Hypothesis 1 (H1):
Beginning second language learners will make use of few constructions – only the copula, intransitive and monotransitive will occur – as the learners become more advanced they will use the ditransitive and the complex-transitive constructions, too; however, the copula, intransitive and monotransitive will be used more frequently than the ditransitive and complex-transitive.
Hypothesis 2 (H2):
Beginning second language learners will use set patterns in constructions, so there is little variation in the use of verbs. As learners become more advanced, they will expand their verb variety. Yet, the few verbs learners used in the beginning for a particular construction will remain to comprise the lion’s share of the total amount of verbs that more advanced learners use.
Hypothesis 3 (H3):
individual learner’s growth patterns shows non-linear development, variation and peaks and regressions.
These hypotheses are discussed by means of the results from the various analyses on the use of constructions, the use of be and have and the use of tenses.
5.1 Use of constructions and verbs
The lower-level students (levels 0 to 2) mainly used the copula, intransitive and monotransitive constructions, with one or two exceptions in which they incorrectly employed a complex-transitive or a ditransitive. From level 3 onwards the ditransitive construction was used correctly and more frequently. However, the complex-transitive construction was hardly made use of even at levels 5 and 6. The copula, intransitive and monotransitive were the most frequently used constructions at each level. Even for the higher levels these three constructions made up the largest share of constructions. All in all, H1 is borne out, because beginning L2ers (i.e. levels 0 to 2) only use copula, intransitive and monotransitive constructions – although there are some incorrect uses, mostly literal translations of Dutch, of the ditransitive and complex-transitive – while intermediate L2ers (i.e. levels 3 and 4) begin to use the other constructions correctly and more advanced L2ers (i.e. levels 5 and 6) use all constructions, yet even in the latter groups the copula, intransitive and monotransitive were used most frequently.
linear as would be predicted by DST. Bumpy growth, or variability, has been observed in other studies (cf. Verspoor, Lowie and van Dijk, 2008). These studies commonly adopt a DST framework in which it is argued that variability is not only due to external sources, but internal reorganisation as well. Verspoor, Lowie and van Dijk conclude: “These microgenetic analyses have shown that the amount of variability constantly changes and that progress and regress follow each other, showing nonlinear patterns of development.” (2008, p. 229). This might explain the small dips in total use of constructions at levels 2 and 3; various strategies might be tried out and the system is constantly adapting and reorganising. All in all, H3 with regard to the use of constructions is only partially borne out, because the use of constructions increases, but the growth is nonlinear and uneven. To determine whether H2 is borne out each construction is discussed separately.
5.1.1 Copula
