Complex Sentences and their Punctuation in English Texts Composed by Latvian Students

Comma use in Latvian and English complex sentences differs significantly and this can cause major problems to Latvian users of English. In particular, the dependent clause that follows the independent clause is normally separated by a comma in Latvian complex sentences (Blinkena, 2009), whereas comma use/non-use in English complex sentences (Downing, Locke, 2006) depends on the semantic relation between the independent and the dependent clause. These comma use differences, as previous research reveals (Farneste, 2006b), can cause challenges in English complex sentence punctuation to students whose native language is Latvian. The goal of this study is to research Latvian students’ punctuation of English complex sentences extracted from a corpus of essays. The analysis covers the cases of nominal and relative dependent clauses that are tightly related to the independent clause they follow. Normally, as Downing and Locke (2006) note, such dependent clauses are not separated by a comma in English texts. The analysis shows that a strong semantic relation between the clauses has been relevantly recognised by students in numerous examples, which they have relevantly punctuated. However, the results also reveal unnecessary separations of relative and nominal clauses from the independent clause by a comma. These findings call for the focusing of students’ attention on the role of the semantic relation of clauses in English complex sentence punctuation by contrasting these sentences with the relevant complex sentences found in Latvian texts. In addition, further analysis of various sentence types and their punctuation in student-composed texts of different communicative purposes is required.


Introduction
Comma use in Latvian and English complex sentences differs significantly and this can cause major problems to Latvian users of English.The dependent clause that follows the independent clause is normally separated by a comma in Latvian complex sentences (Blinkena, 2009), whereas comma use/non-use in English complex sentences (Downing and Locke 2006) is related to the semantic interdependency between the independent and the dependent clause.These and other differences in comma use have been previously discussed by Farneste (2006aFarneste ( , 2006b)).She concludes that differences can cause English sentence punctuation challenges to students whose native language is Latvian.Farneste (2006a, 37) emphasises that first-year university students in Latvia, when asked if they discussed punctuation in English texts in English classes at their schools, predominantly point out that it was rarely done.The author also notes that there are very few tasks in English course books focusing on the communicative function of punctuation marks.Therefore, students often tend to rely on the punctuation rules that have been discussed in Latvian language classes.Meanwhile, the research on comma overuse in complex sentences in English texts written by Latvian students has not been taken up yet.
The goal of this study is to explore examples of complex sentences written by Latvian students in order to uncover how far their punctuation corresponds to the semantic relation between the independent and dependent clauses.
Seeking to achieve this goal, the following enabling objectives are set: (1) a comparative analysis of the theories on sentence classification according to clause relations and clause types; (2) extraction of examples of complex sentences from a corpus of essays composed by Latvian students (with the help of AntConc) that have been part-of-speech tagged with CLAWS7; (3) grouping of the extracted examples into sentences that contain nominal and relative dependent clauses; (4) analysis of the semantic relation of the dependent and independent clauses and their punctuation as proposed by the students.

Theoretical background
Sentences, 'the highest grammatical units' (Downing, Locke, 2006, 272), that contain more than one clause or verb phrase (Biber et al., 1999, 100, 120) as a central element are classified by linguists in terms of the relationship of the elements within the sentence.Researchers of the Latvian language (e.g.Auziņa et al., 2013, 831;Beitiņa, 2009, 173;Ceplītis et al., 1989, 110-111) label all such sentences as composite sentences (in the Latvian language: salikti teikumi) and use this as an umbrella term for all sentences that go beyond one clause.This term is not normally found in the grammar of the English language.However, it has been used by some linguists, for example, Lyons (1995, 157) in order to distinguish 'non-simple' sentences, which contain more than one clause, from simple sentences.It is also used by Davidson (2005, 30) in his punctuation manual 'How to Punctuate'.
Apart from compound and complex sentences, the so-called compound-complex sentences (in the Latvian language: jaukti salikti teikumi) are distinguished by the researchers of the Latvian language.They contain several clauses of equal as well as unequal status (e.g.see Auziņa et al., 2013, 831).In this connection Downing, Locke (2006, 273-174) explain that coordination and subordination of clauses in a connected discourse 'may be […] complex and variable', so they expand the notion of a complex sentence by claiming that 'a complex sentence can consist of any number of clauses of different types and in different combinations'.
Complex sentences or sentences with unequal relation of clauses are further described by linguists from two angles -the type of dependent clause and the semantic degree of subordination in its relation to the independent clause, both of which play an important role in punctuation solutions in English.According to traditional grammar, there are three main types of dependent clauses that are named after the relevant parts of speech: Noun, Adjective and Adverb clauses; however, there are terminological variations in the description of Noun and Adjective clauses.Noun clauses (Yule, 2006, 160) are referred to as nominal clauses by Biber et al. (1999, 193) or complement clauses by Trask (1993, 10) and Hurford (1994, 232), for example.The term 'adjective clause' or 'adjectival clause' seems to have been abandoned by linguists in favour of the term 'relative clause' (Biber et al., 1999;Downing, Locke, 2006;Yule, 2006, etc.).The term 'adverbial clauses' is generally accepted and has been widely used by linguists.
Although the previously mentioned dependent clauses are in a subordinate relation to the relevant independent clauses, the degree of subordination of dependent clauses Didaktinė lingvistika can vary in a complex sentence.Downing, Locke (2009, 275) emphasise that it is important to '[…] adopt the view that dependency is not an absolute property, but rather a question of degree'.They explain that the dependency degree shows the degree of integration of a dependent clause as 'perceived or imagined by the speaker or writer between events'.The tightest integration is referred to as embedding by linguists, which means that the dependent clause functions as a constituent of the independent clause.
On the basis of the aforesaid classification of complex sentences, some linguists point out that the English dependent clauses, which are embedded within complex sentences (e.g.Downing and Locke, 2009) and, therefore, do not require separation by a comma, are: nominal that-clauses (Example 1); nominal wh-interrogative clauses (Example 2) including whether and if, as well as relative (restrictive) that and wh-clauses (Examples 3 and 4).
(1) He saw that the bottles were empty.
(2) I'll ask where the nearest underground station is.
(3) The house that they live in is large.(4) Perhaps the people who were waiting are still there.
The perception of the degree of dependent clause subordination plays an important role in complex sentence punctuation in English.As it is explained in English punctuation manuals, a comma is used 'to make a slight break' (Davidson, 2005, 78), and for this function Lukeman (2007, 31) has metaphorically described a comma as 'the speed bump of the punctuation world' because 'with its power to pause, the comma controls the ebb and flow of a sentence'.This implies that the pause marked by a comma is relevant in the cases where the dependent clause is less tightly integrated.Comma variation caused by semantic dependency in English complex sentences obviously explains why English punctuation is considered a challenging issue of text composition by Latvian students of English (Farneste, 2006a(Farneste, , 2006b)).
The interrelation of an embedded subordinate clause separated by a comma reveals a noticeable, though neglected, aspect of comma use differences in Latvian and English complex sentences.Although some researchers discuss the semantic relation of clauses in Latvian complex sentences (e.g.Beitiņa, 2009), this phenomenon does not have the same impact on the punctuation in Latvian texts as it has in texts written in English.In Latvian complex sentences, as Blinkena (2009, 176) states, all dependent clauses, irrespective of their semantic dependency degree, are normally separated by a comma (see Example 5 and its translation into the English language -Example 6).
Thus, the theoretical analysis has revealed that comma use/non-use in English complex sentences, containing an independent clause followed by a dependent clause, is based on the type (e.g.nominal, relative) and semantic degree of subordination.The neglect of semantic dependency might cause punctuation confusion as, for example, in the following case: a nominal who-interrogative clause (Example 7), which is embedded in the main clause and, therefore, is not separated by a comma, might be confused with a non-embedded (i.e.non-restrictive/non-defi ning) relative clause introduced by who (see example 8) and, therefore, separated by a comma.The loose dependency degree of all the wh-non-restrictive/non-defi ning relative clauses should also be marked by the comma use.(7) They know who has left this paper on the table.(8) I'll give the CD to Ben, who likes music.

Method, procedure and results
The material for analysis was extracted from a Latvian student-composed text corpus and the analysed set contained 104 texts of approximately 21,319 words.The texts were written by students before their participation in a writing course.Their selection aimed at revealing if the students had used the above-discussed clause types and patterns, namely, the embedded clauses (nominal and relative) that are preceded by an independent clause, and what punctuation solutions they offered to serve as a departure point for the development of their writing skills.
In order to extract these clause types and patterns, corpus-based data extraction procedures were applied (McEnery et al., 2006;McEnery, Hardie, 2012).The clauses were extracted with the help of the AntConc concordancer.Seeking to ease data extraction, the texts were part-of-speech tagged by CLAWS7, which contains tags for the linguistic features functioning as subordinating conjunctions in complex sentences that are in the focus of this study (see Table 1).The tag CST of CLAWS7 was used as a query to extract nominal and relative that-clauses as conjunction concordance lines that were sorted into nominal and relative The students' texts, as expected, contained both nominal and relative that-clauses.The majority of these clauses (Table 2) were correctly presented by the students as embedded, as they had not been separated by a comma.Example 8 shows an embedded nominal clause, and Example 9 an embedded relative clause that, due to their strong semantic relation, are not separated by a comma.The students embedded their nominal clauses in such common structures as I think; I would like to tell; people think; I knew that; I must say; I can conclude' etc., as seen in Examples 8 and 10. (8) Some people think that travelling is a bad and useless way of spending money.(9) There are many routes that go through the city.
There were also occasional dependent nominal and relative that-clauses that, irrespective of their embedded information, were separated by a comma.The two sentences in Example 10 show comma use in nominal clauses, but example 11 shows comma separation of a restrictive relative clause.(10) I believe, that this way of transport is quite developed.// I think, that it is a beautiful city.(11) It has been circumstances, that are the most negative factor.
The overall percentage of nominal and relative clauses that were/were not separated by a comma was equal.It has to be noted that commas would normally be used in such sentences in the Latvian language, so the redundant comma use might have occurred due to the impact of their use in the Latvian language.The extraction of nominal interrogative clauses required the application of several tags.The tag RRQ extracted wh-general adverbs and the output included not only nominal wh-interrogative clauses, but also e.g.direct questions and adverbial clauses.Therefore, in order to extract nominal clauses that stand after an independent clause, the search was narrowed to VV* * RRQ and other tag modifications (Table 1).They were all placed in the advanced search option, as well as the output concordance lines were examined and sorted to check each instance for relevance.Presuming that the search options had enabled the extraction of all nominal interrogative clauses found in the texts, the results showed that students had used a small number of them.Although the majority of the clauses were not separated by a comma (as in Example 12), there were also cases where the students had separated these clauses by a comma, as in Example 13, where the author might have intended to insert a pause after 'wonder' for a dramatic effect.Obviously a dash would have been more relevant in that case.(12) I didn't know where they were.(13) One might wonder, what will happen.
It has to be noted that the number of these clauses extracted is too modest to serve for generalisations.
Although the PNQS tag extracted who in different position within a sentence, e.g. at the beginning of sentences, which were just a few cases in the students' texts, it allowed convenient extraction of relative who-clauses, which in their turn could be sorted into restrictive and non-restrictive clauses to check the relevance of comma use.Relative who-restrictive/embedded clauses that are normally used without a comma can cause punctuation challenges, as who is used to introduce both restrictive /embedded and non-restrictive/non-embedded clauses.Indeed, who relative restrictive clauses turned out to be challenging.Apart from numerous relevantly punctuated instances (Examples 14), there were also cases of the irrelevant insertion of commas (Examples 15).( 14) Of course there are people who mostly prefer walking.
There are a lot of people who are driving to work by car.
People who do not have their own car.(15) There is anybody, who would love.
I feel it more than people, who live in a city.( 16) Now factories who make cars think how to make their cars friendly to nature.
In addition, as can be seen in Example 16, there are clauses in which the use of who is doubtful, as who wrongly refers to factories, though it normally relates to humans.It is also noteworthy that cases of non-restrictive who relative clauses were not found in students' texts, which means that they did not display varied approaches to sentence structure to reach the communicative purpose of the text.
The PNQ0 tag was employed to check if the cases of nominal and relative whom clauses were found in the texts.It was determined that these clauses had not been used by the students.