turkishenglish.com - Comparison of relative clauses in ESL students



The Korean students use SS relatives (Subject Head with Subject NPRel) more frequently than the Spanish students and SO (Subject Head with Object NPRel) relatives are extremely rare, comprising only 5.7% of the RCs with Subject Heads.  The significance of this can be seen if we look at the distribution in Fox and Thompson’s data.  A Subject Head receives an Object relative 50% of the time.  Because of the flow of information the Subject is frequently ungrounded when the speaker hears it and so, in English, an Object relative is used to ground the Head NP by anchoring it to another NP, usually a pronominal form.  This is not the case in Korean and Japanese.  Grounding is not one of the major roles of the RC in these languages, and so the large number of SS relatives reflects the propensity of their native language to use more Subject Heads than English and to ground those Heads not by anchoring but by characterizing them, which produces an S relative.  The Spanish students, on the other hand, match native speaker distribution rather well. 

            To determine what sort of discourse function these SO RCs might be performing other than grounding, I examined every occurrence and found this combination of S Head with O NPRel can perform at least two functions. 

            The first function is that of presenting a new sub-topic in terms of the overall purpose of the essay or paragraph.  I call this “topical cohesion” since the writer’s purpose seems to be to remind us that the new sub-topic is directly related to the essay’s macro-purpose.  The English RC is perfectly suited to this function (cf. Givon 1993).  In (11) below, the student is writing about what she likes to do.  She is in the middle of the paragraph and the first lines of the paragraph have simply been a list of what she likes to do.  Then, she uses an SO RC to introduce the main point of the paragraph. 


            (11)      I like to swim, to run, and play tennis,  I think that do excersise is very   importat to the health.  I know play guitar and I want to learn play piano next           year.  But definitly the think that I most enjoy is travel.  I like travel very             much,   I hope to travel around in the world because I think that is very interesting for me to know persons who speak others language and who has       others custom.  (S3.1.2)


In the rest of the paragraph, the student continues to write about traveling.  She has used the RC that I most enjoy to frame the new sub-topic in terms of the paragraph’s main idea.  Six of the fourteen SO RCs used by the Spanish students were used in this way, starting with the level 3 students.  Another example is (12).


            (12)      The third thing that I need to acquire my happiness is by being a good person in all the stages of my life.  (S 5.16.10)


Here, the student is writing about the requirements for happiness and introduces a new sub-topic, "being a good person," within that frame. 

            One of the two Korean SO RCs is of this type.  It was produced by an advanced level student, and does a wonderful job of framing the new sub-topic. 



            (13)      During the 1970's, the Korean government started the conservation of   nature movement because the mountains were not green but brown, and the rivers were not blue but gray.  They just concerned with industrial progress after Korean war.  And they did not realize that the nature was destroyed by the industrial progress before the 1970's.  Thus they set up some plans to protect nature during the next two decades.

                        The first thing that the government prepared was the department of conservation of nature which was concerned with all things about nature. 


This student introduces a new sub-topic in terms of the plans of the government.  The entire paragraph introduced by the RC is about the department of conservation of nature.  Let me give some examples of Korean students' presentative constructions that lend themselves to the use of an RC, but where they did not use one.   All of the following examples except the last, which is on the differences between generations,  are from essays written on the requirements for happiness.   All the examples are the first sentence in an orthographic paragraph.


            (14)      Therefore, first important thing is to control mind by themselves. (K3.2)

            (15)      The second requirement is money.  (K3.6)

            (16)      Health is the most important pre-condition for happiness.  (K3.5)

            (17)      First we will be happy if we possess something.  (K3.24)         

            (18)      The first important requirement for happiness is optimism.

            (19)      The first different things begins the background of birth.

                        The second thing is the purpose of life.

                        The third thing will "the education of children." (K4.16)


            The first two examples really have no topical cohesion at all.  In (15), for instance, the reader is forced to infer that money is related to happiness.  We expect some sort of restrictive device, either an RC or a preposition.  Example 15 might have been rewritten “The second requirement for happiness is money.”  Examples 16-18 do provide some framing of the new sub-topic, but not with an RC.  The last example (19) is particularly interesting because the Spanish students frequently use thing as a Head Noun when framing a new sub-topic and this Korean student does the same, but without any type of post-modification specifying what type of thing he is writing about.  He introduces all of the main points of his essay without any type of topical cohesion.  Again, we have to infer the realtionship, which, one must admit, is clear in the context of the essay, but a native English speaker expects it to be made explicit.  

As one can see,  the Korean students rarely use RCs to frame sub-topics, and in fact many of their new sub-topics receive no framing at all.  But, Spanish students often used RCs to introduce a new sub-topic in terms of the essay guiding premise, and not only SO RCs but SS and SA RCs as well. 


            (20)  The other thing that could help in feeling happiness is having all the    basic things that you need in life.  (S 3.13.13)


This difference in discourse presentation accounts for part of the difference in

production between the two groups.  The Spanish students are clearly using RCs in conjunction with Subject Heads to perform a specific discourse function that the Korean students do not. 

            The second function of the SO RC is similar to the first in that it serves a topicalizing function even in terms of the essay’s purpose, but no new sub-topic is being introduced, and the information in the RC is script predictable.  For these reasons I have considered these examples different from the first ones.   In the next examples, the students are writing a compare and/or contrast essay on the generation gap.  Both sentences are taken from the final paragraph of the essay.


            (21)  The comparison that I can make is that I am studying for be something             in life, to be successful as my father did.  (S 4.1.5)


            (22)  The only thing that I want to demonstrate is that there is a generation gap in everywhere you go.  (S 5.10.1)


Both of these RCs allow for cohesion, but they are not used to frame a new sub-topic, rather they are used to frame a summary or concluding statement in terms of the student’s own present activity, making a comparison and demonstrating something. 4    The other Korean SO RC is similar to this.  Not that the student frames a discussion in terms of her own activity, this is a purely Spanish phenomenon, but that she uses redundant information to produce a frame.  The student is writing about requirements for happiness and, in the paragraph the example is taken from, she is writing about the relationship between happiness and success in whatever we attempt.  Note the use of the noun undertaking. 


            (23)      Second, when they succeed in their life, they will feel happily.  It is a big happiness to succeed in their undertaking or pass their test... And if everything they do prospered with them, then they may be happy.



The Head Noun and RC are equivalent to a semantic “all your actions”, and when coupled with the matrix verb prosper, we have success. 

            There were two examples of SO RCs that did not seem to fit into these two categories.  One of them was a straightforward case of grounding.5  The other example added background material.6

            As with SO RCs, there are some definite discourse functions performed by Predicate Nominals with RCs.  The Korean students tend not to use this construction and the attending functions, having only 11 P Heads while the Spanish students had 53.  I divided the RCs with P Heads into 4 categories: presentative, topical cohesion, topic maintenance, and descriptive.  These categories are not mutually exclusive. 

            The first function is similar to that of the SO RCs discussed above.  Relative clauses with Predicate constructions were used fourteen times by the Spanish students to introduce a new sub-topic and 8 of these occur with an RC which provides topical cohesion.  In the following examples the students were writing about the causes of happiness and they start a new paragraph with a Predicate Nominal construction which is used to introduce the next factor in attaining happiness. 


            (24) Success in life it's an other factor that cause happiness.  (S 4.2.5)

            (25)  Being healthful is one of the most important aspects that can give you happiness.  (S 3.13.6)


            This presentative function did not always overlap with the framing function of the RC.  Sometimes, as in (26) below, the RC provided background information characterizing the topic or providing grounding and sometimes the RC was the new topic and the main clause provided the grounding, as in (27).


            (26)  Learning English as a second language is something that open you mind, that help you to meet better other people's ideas.  ( S5. 7.1-2)

            (27)  The main similarity between Arlington and Medellin is the way that the people are with the other persons.  (S 3.8.3)


            The Spanish students seem to have two different strategies for introducing a new sub-topic and making an explicit link to the essay’s theme.  One is the RC with a Subject Head and the other is a Predicate structure.  However, they locate the topic in different parts of the sentence.  The Subject Head structure places the new sub-topic in the predicate, and the Predicate structure places it in the subject position.



            (28)  Another factor that causes happiness is success in life.

            (29)  Success in life is another factor that causes happiness.


There seems to be a tension between the two.  In (29) the topic is in subject position, the prototypical grammatical relation for a topic.  But, it is a new sub-topic and new information is usually in the predicate.  This status of being not really one nor fully the other may be the reason that there is a choice in English as to how to present a new topic.   

            Another function of the Predicate Nominal constructions with RCs seemed to be that of keeping the current topic in the Subject position and allowing the RC to make an assertion about the topic.  I have called this “topic maintenance”, and there were 10 examples of this.  Example (30) illustrates how this works.  The student is describing Monterrey and later in the essay compares it with another city, Reynosa.



            (30) I will talk about Monterry first.

                        The people called it "the city of the industries," because it has a lot of them, the city receive a lot of bussiness man who looks for increase their market;  Monterry, Mexico and Guadalajara are the citys where the majority of mexicans are living.  The Monterry city is not a tourist city, ...             (S3.5.2)


            The use of this construction may be due to the word order of English.  All of the examples of this usage are PO and PX RCs.  An O or X relative locates the referent in the predicate of the sentence, which is usually the locus of new information.  If the referent is the essay topic, this is inappropriate.  The shift to a Predicate construction allows the topic to continue to be in the subject position and the P itself is a Generic Head, such as thing, something, place, factor, aspect, etc.  The sentence final RC then makes the assertion about the topic.  If these sentences are turned into canonical sentences without RCs, which can be done in every case without loss of meaning, then the essay topic is forced out of the Subject position and into the Predicate.  Compare (30) and (31) and observe how the lack of a Predicate construct interrupts the flow of information in (31).  


            (31)   I will talk about Monterry first. The people called it "the city of the industries," because it has a lot of them, the city receive a lot of bussiness man who looks for increase their market;  The majority of Mexicans live in Monterrey and Guadalajara. 


The Korean students use only 4 PO and PX RCs, and though 3 of them seem to function in this manner, the problem is that they use them much more infrequently than the Spanish students. 

            The other function of RCs with Predicate Nominal Heads is that of description.  Of the 22 descriptive RCs with Predicate Heads, 9 of the RCs are grammatically unnecessary.   That is, the Predicate structure is not being used to allow for additional information.  Some simple examples should clarify what I mean.  In (32), the information in the RC can be encoded with a matrix clause without any loss of meaning, being adequately replaced with (32a).


            (32)  God is something very important who gives me power, hope, patient.  He is the one who gides my life.  (S5.13.3)

            (32a)  He guides my life. 


Even though the RC in (32a) does an adequate job of replacing the Predicate structure semantically, it seems to be pragmatically inappropriate.  The Predicate structure in (32) flows much more smoothly than the simple matrix clause in (32a), which seems more abrupt and weaker.  Why should this be so?  This structure seems to focus the information to highlight its importance, by repeating the noun in the Predicate.7  Only one of the 11 Predicate structures used by the Korean students could be reduced to just one clause.  They do not seem to use PNs to focus information, and consequently do not produce RCs in this way.  Note that the information in the RC of (33) cannot be moved into the matrix clause.


            (33)  In conclusion, Sadang and Ynyang are both good places which is easy to live.  (K3.7.1)

            (33a)  ?* Sadang and Ynyang, both good places, are easy to live in. 

            (33b)  Sadang and Ynyang are good places and they are easy to live in.


As (33b) demonstrates, the reason that the RC cannot be changed into a matrix clause is because there are really two predications involved here, hence the need for an RC to make an additional assertion.  

            Other Heads also had different NPRel distributions.  The Spanish usage reflected that of the native speaker in every category except XO and XX (see Tables 21 and 23).  The native speakers use more Object relatives than X relatives on X Heads.  The Spanish students used an equal number of both.  This difference may turn out to be quite insignificant because by far the majority of X relatives are OPREP and though differing from O relatives, they are still suited for  similar functions.  Compare,  “Have you been in the new house Bob bought?” with “Have you been in the new house Bob lives in?”  Both relate the Head NP to some referent.  The Korean students tended to make these S relatives.  This is consistent with the tendency for Koreans to make any relative an S relative.  


Information Status of the Relative Clause:

            The second difference between the two groups was the information status of the RC.  The Spanish students tended to use RCs to identify referents much more frequently (50% more) than the Korean students.  Remember an RC was considered identificatory if both the Head NP and the content of the RC are old information.  In (34), the writer is discussing the difference between Caracas and San Antonio.  A relative clause is frequently used to facilitate the reader in keeping track of what city is being discussed.


            (34)  One big difference between Caracas and San Antonio is the weather.  In   San Antonio, very often it is rain.  The sky regularly is gray in the afternoon. People who live in San Antonio can see the sun just in the mornings.  In contrast, Caracas has...  Another difference between Caracas and San Antonio is the night life.  People who live in San Antonio  go to bed earlier than... (S3.8.3)

This is a big difference in the number of Relative clauses used for some specific discourse function and accounts for some of the difference in total production. 



Table 24.  Information status of RCs in the two groups. 


Type of RC

Language Background




215  (76.0%)

68  (24.0%)


119  (83.8%)

23  (16.2%)


Macaulay (1990) gives us some indication as to which of these groups is closer to the native speaker norm in this regard, and it is the Korean group.  In her analysis of native text, she found that only 7% of the RCs in English used old information to identify a Given referent.  Native speakers can be expected to be more skillful and efficient in their construction of a text and in making sure that the referents in a text can be identified by the reader.  Native speakers do not seem to resort to the use of RCs for this purpose very frequently. 


The Presuppositionality of the Relative Clause:

            The third difference between the two groups is the presuppositionality of the RC.  Of the informative RCs used by the two groups, the Korean students had a larger percentage of RCs which presented non-presupposed information.  Example (35) is presupposed while (36) is asserted. 


            (35)  I try to make them bigger and better than the life that I have seen. (K6.9.1)

            (36)  The other is the positive attitude which emphasizes that the nature must

            be protected.  (K 6.2.1.)


Table 25.  The presuppositionality of the RCs in the two groups. 


Type of RC

Language Background




88     (40.9%)

127     (59.1%)


58     (48.7%)

61       (51.3%)


There are several factors which are related to whether an RC is presupposed or not, namely: the type of matrix verb, the type of Head NP, the Definiteness of the Head NP, and the content of the RC.  We will discuss each of these factors and how the students differed.

            Fox and Thompson (1990) discussed the frequency of the semantically neutral have as the verb of the main clause.  They argued that when an RC is used with this verb, the main clause does the grounding and the RC is free to make assertions.  This verb actually has a sort of dual function in English.  Sometimes it denotes strict possession, as in "I have two cars that I could sell you."  Other times this possessive focus seems to be diluted and the verb is used to introduce a new referrent, as in "I have an uncle who likes to play cards."  In English there is some overlap between these functions. 

            This brings us to the next difference between the two groups.  The Koreans tend to use have as the main verb more frequently than the Spanish students do.  This is part of the reason for the higher percentage of asserted RCs in the Korean data.   Remember the number of matrix clauses with have dropped for both groups as proficiency increased, but the level 6 Korean students used it 7.4% of the time, and this is about the percentage at which the level 3 Spanish students began (8.8%).  The Spanish students in level 6 only used it 2% of the time. 



Table 26.  Occurrence of have as the matrix clause verb. 





have as main verb

12     (4.3%)

16     (11.3%)


The frequency with which the Korean students use have is surprising since it is not a presentative structure in their language like it frequently is in English.  For example, the sentence "I have a friend who is living in Japan" is not the way a new referent is introduced, (Hwang personal communication).  Korean typically uses an existential to perform this presentative function. 


            (37)  To me (there) exists a friend who is living in Japan.


The Korean students seem to have incorporated this form/funtion correspondence into their L2 competency quite readily, but not without difficulty.  As we noted earlier when discussing hypothesis two, their use of RCs with have is sometimes strange, (see examples 1-3). 

            Another major difference contributing to the difference in the status of the RC between the two languages was the type of Head Noun.  The Spanish students often used Indefinite pronouns, such as someone, something, somebody, everyone, etc., and what I have termed Filler Heads, such as thing(s).  The Korean students used these forms very infrequently.  The percentage of indefinite pronouns used by the Spanish students was over double that used by the Korean students, and the percentage of the Filler Head NP, thing(s), was almost 7x that of the Korean students.  This difference is significant since Head NPs of this type are, as I have demonstrated earlier, usually presupposed.  In fact, they can frequently be replaced by a complement clause functioning as an argument of the matrix verb.  Thirty-seven of the RCs produced by the Spanish students could be replaced with a complement clause while only 5 of the Korean RCs could have a complement clause substitute.  Actually (38) and (38a) are the same in Korean.8


            (38)  The thing I enjoy most is traveling.

            (38a)  What I enjoy most is traveling. 


            The frequent use of such Head NPs also explains part of the difference in total production between the two groups since Head NPs of this type practically require modification because of their indeterminate status for the reader.  In these sentences the reader often has to ask himself, "What kind of thing is the writer referring to?"  The answer, of course, is often provided by a relative clause. 

Table 27.  Distribution of some of the Head NP types. 


Lang. Background

Type of Head NP



Indefinite PRO

42     (14.9%)

10       (7.0%)

Filler Head thing

39     (13.8%)

  3       (2.0%)


  9     (  3.2%)



  6     (  2.1%)

  1       (0.7%)


96     (33.9%)

14       (9.9%)


            The Korean students show a strong tendency not to use such NPs as the Head of an RC in English, but all of these Filler Heads are common in Korean, and one of these, time, is used with an RC construction in all cases where English uses a temporal when clause (Hwang 1993). 


                                     [RC]               T                    [RC]               T/S1

            (39) [hakkyo-lo     ka-l]           tta-pota-nun     [o-l]              tta-ka         amulato               school-to      go-fuM     time-than-Top come-fuM  time-NOM  after.all

                  S2                  V    

                maum-i          nohi-nun               moyang            ita

                mind-NOM   be.at.ease-prM      appearance       is

                It appears that, after all, minds are more at ease when coming from school

                than when going to school.  (Hwang 1993:7)


            The Korean students have not transfered this function of limiting a generic head with an RC in their attempt to learn English.  This could be interpreted as good teaching by someone in their past learning experience, but the fact that English does not normally use temporal nouns as the Head of an RC is not something that most ESL teachers are probably aware of, and the fact that it is sometimes possible in English would suggest that if anything they might be taught how to do it instead of how to avoid it. 

            Notice also that none of the Korean students used way as a Head NP.  Sometimes their use of way is somewhat peculiar because they tend to use a preposition to restrict the reference of the Head.  Example (40) was tested on several native speakers who all thought it a bit odd.


            (40)  The way of thinking is different depend on people.  (K6.7)


            The fact that so many of these RCs with Filler Heads can be replaced with complement clauses makes one question the use of such constructions by the Spanish students.  Spanish students may be avoiding the complement clause.  A count of the frequency in discourse would be needed to answer these questions and might be an area for further research.  As to the lack of Filler Heads in the Korean data, it may be that the Korean students are wary of using RCs in English to encode information that is an RC in their native language because of the perceived distance between the two languages and because of prior difficulty in transferring what would be RCs in their native language to English (cf. Kellerman (1977) on other factors affecting avoidance).

            There is another difference between the two groups in reference to the type of Head NP.  The Korean students had a tendency to use personal pronouns as Heads, even in the intermediate levels.  This had stopped entirely by the time they were considered advanced students.  The Spanish students never used personal pronouns as Head NPs.

            The Definiteness of the Head Noun is also a factor in the number of relative clauses which are asserted rather than presupposed.  Indefinite NPs are more likely to have non-presupposed information in their RCs than are definite NPs.  If we take just a superficial glance at the numbers, they would seem to contradict the figures already given in Table 25 because the Spanish students have a higher percentage of Indefinite Heads than the Koreans do. 

Table 28.  Definiteness of Head NP.






129     (45.7%)

153     (54.3%)


  69     (48.9%)

  72     (51.1%)


The question we want to ask is why, if they have more indefinite Heads, do the Spanish students have fewer asserted RCs.  This can be accounted for if we remember that the Spanish students often used Indefinite PRNs and the Filler Head, thing, which was often indefinite.  These constructions were most frequently presupposed even though indefinite because they resemble complement clauses in that the Head NP is subsumed and it fails the separability test.  Adjusting for these by subtracting them from both groups gives us the following figures, where the Spanish students now have a higher percentage of DEF Heads than the Koreans.   



Table 29.  Definiteness of Head NP after adjusting for Indefinite PRN and thing.



Definite Heads

Indefinite Heads


129    (57.3%)

96       (42.7%)


69      (53.5%)

60       (46.5%)


            The issue of presuppositionality brings up another difference between the two groups.  The Spanish students frequently used RCs to encode logically redundant information or RCs whose only function seems to be that of grounding.  The information encoded in these RCs is always presupposed.  The first two examples below are of the first type.


            (41)  About the food, I can tell you that it is really different in these places which I am talking about.  (S 6.3.7)

            (42)  The only thing that I want to demonstrate is that there is a generation gap

            in everywhere that you go.

All of these examples have non-human Heads with O or X relatives and pronominal subjects in the relative clause.  Example (41) above is taken from the end of an essay comparing Caracas, Columbia and Radford, VA, so the RC is extremely redundant.  The reader has known that the writer has been talking about these places for some time now.  The entire RC in (42) could be omitted.  The information is unncessary, but it sounds like perfect English.  Maybe both of these are examples of the explicitness with which the Spanish students approach writing.  We saw something akin to this in that they used far more identificatory RCs than the Korean students. 

            The next examples are RCs with have in the RC.  Fox and Thompson (1990) gave this example (43) of a typical grounding RC.  This could easily have been a possessive, as in (43a)


            (43)  Well see what the problem I have is my skin is oily and that lint just flies     into my face.  (Fox and Thompson 1990:303)

            (43a)  Well see my problem is my skin is oily...


The relative clause in (43) can be a possessive because it is an O relative with a pronominal subject.  The Spanish students used 19 RCs with the verb have of which 13 were O relatives.  The Korean students used 15 RCs with have, but only one was an O relative.  Example (44) is taken from a Spanish student’s composition.


            (44)  In my opinion, Mr. Salinas has been the best President that Mexico has had.  (S4.10.8)


            The last difference between the two groups was the number of NRRCs used by the two groups.  This is related to the issue of presupposition because NRRCs are never presupposed.  As expected from what I know of relativization in Korean, the Korean students used far more NRRCs than the Spanish students.  The Korean students had 32 NRRCs accounting for 24.6% of the total number of RCs.  The Spanish students had 27 NRRCs which made up 9.5% of their total and was less than half of the Korean percentage. 

            In conclusion, the hypothesis was strongly supported in a number of areas.  The purpose of this thesis was not to prove transfer, but rather to account for the difference in production in terms of IL variances.  This was accomplished successfully, but it was pointed out whenever a particular aspect of the IL seemed to be due to the influence of the NL.  I will briefly highlight some of the more salient differences between the two groups.  

            The distribution of the GR of the NPRel and the Head NP showed substantial variance with the Korean students clearly transferring their native distribution to their English writing.  There were some discourse functions which were associated with the areas in which Korean and Spanish students demonstrated the largest variances.  The Spanish students also used significantly more RCs for identification than the Korean students.  The type of Head Noun was another difference between the two groups.  The Spanish students used far more Filler Heads than the Korean students.  The Korean students used more NRRCs than the Spanish students and this is also probably a result of transfer.  All in all, there was positive evidence that the Korean students use RCs in substantially different ways than Spanish students do, and that a large part of the total difference in production can be explained in these terms.  

Hypothesis: #4  The Koreans will have more instances of negative transfer with respect to the function of the RC than the Spanish students.

            This hypothesis was confirmed.  There was one particular area in which the Korean students used RCs to perform a clearly non-native function.  After making reference to some group or class, they often used an RC to encode a set-some-member relationship.  There were nine examples of this type of RC.  All of the occurrences were in the intermediate level compositions.  The students had stopped using the RC for this purpose by the time they reached the advanced levels.   


            (45)  If some couple try to get married, they may be needed to a lot of things which are articles essential to marriage, their parents permission, and so on. 


            (46) Firstly, we are living differently each others in the world.  Someone’s target should be doctor, and somebody’s target may be sportsman that is a like baseball, football, or basketball.  (K3.22.3)


Both of these examples should use either for example or such as to make the relationship clear.  Hwang (personal communication) informs me that these are RCs in Korean.  Because of the frequency with which they use RCs to perform this function, one can be fairly confident that this is a reflection of their IL competence not just a mistake.  Example (47) is a Korean token of such a construction.  Note the use of the verb kath ‘be.like’ which is exactly like (46) above.  The student seems to have transferred all of the semantic properties of the NL to the corresponding lexical forms in the TL.  There was one other RC that used like.


            (47)  kiph-un           sam-e-nun   [thokki, sasum,  you     kath-un]

                    be.deep-prM   mt.-at-top     rabbit    deer      fox     be.like-prM


                     jimsung-i         mani             it-ta

                     animal-Nom    abundantly   exist-Decl.

                    ‘In a deep mountain, there are many animals like rabbits, deer, and foxes.’


            Actually, a similar construct is possible in English if it is a set-all-member relationship.  Several Korean students used examples like the following:


            (48)  Seoul has four seasons which spring, summer, fall, and winter, Arlington alike that.  (K3.10.4)


These constructions, though often appositives in English, can be non-restrictive relative clauses.  It seems that the first ones are deviant because of the equative relationship which forces us to interpret the RC as being exactly equal to the Head Noun when in fact it is not. 

            I have already discussed the deviancy of the Korean students with respect to relativizing on Personal Pronouns.  This is area in which they manifest a non-native function of the RC in their IL.  There were six examples of Personal PROs as Head NPs. 

            There were 8 other RCs which were considered deviant in function.  There was not any one characteristic that they all had in common, so there is no category for these deviancies.  I interpret some of these to be part of the students’ struggle to understand exactly how and when to use English RCs.  Some of them have already been discussed, such as the use of an RC to report redundant information when the verb of the matrix clause is have (see examples 1-3).  Another is the incorrect use of NRRCs.  In (49), the information in the RC is far too important to be in an NRRC,  and yet the Head NP forces a nonrestrictive interpretation. 


            (49)  Happiness that gives you pleasure and comfort has many causes.                             (K4.19.1)


In (50), the RC should be a participle as in (51).


            (50)  When I came here, I couldn’t see people who walked on the street, but I saw many cars.  (K4.4.2)

            (51)  When I came here, I couldn’t see people walking on the street, ...


            In all, there were 23 RCs in the compositions written by the Korean students which were considered functionally deviant.  This accounts for 16.2% of their total production.  When we couple this with their already low production, we begin to see the seriousness of the problem these students have with the acquisition of English RCs. 

            The Spanish speakers, on the other hand, had only 2 RCs, .7% of the total, which could be considered deviant.  One of them, (52),  seems to be similar to the Korean RCs used for the set-some-member relationship. 


            (52)  Because for example, to some occidental people the happiness is TO HAVE something, like a good job to get a lot of money, to buy a house, a     car, etc. and always want to have more than someone who can be his best friend or maybe his brother or sister, anyway.  (S3.10.3)


The other deviant RC is nonrestrictive.  The entire essay has been about Columbus, so the use of an NRRC to identify him is unnecessary.  


            (53)  In conclusion, this famous person that was Christopher Columbus made one of the most important contribution to the world.  (S 4.11.6)


            Even though the number of deviant RCs in the Korean data is relatively high, I was surprised at the infrequency of RCs and the type of deviancies in the data base.  In my classes,  Korean students seem to use more RCs in their compositions than they did in the compositions I analyzed, and there were definitely more RCs used in a non-native fashion.  I would like to give some typical examples taken from my level 3 class this semester, and comment on why there should be a difference.  One of my Korean students this semester wrote an essay comparing housing in Korea and America.  In this essay he used 6 RCs, which is quite above the norm for level 3 students in this study.  Four of these were semantically NRRCs and 2 were functionally deviant.  The following paragraph is the introductory paragraph and neither of the two RCs in this paragraph were offset with commas, but they are clearly nonrestrictive. 


            (54)      It is necessary that human beings are equipped with three elements; food, clothing, and house.  These are inseparable from one another in our life. Of all the elements, house might be the most important element (a) that gives the  place to sleep and enjoy ourselves.  (b) The house which is influenced by the evironment can be said to have a big difference between Korean and America.  (K3)


To interpret (54a) as restrictive forces us to conclude that our house is just one of many places we sleep and enjoy ourselves.  Actually, the information in the RC gives the reason for the statement in the matrix clause.  After I conferred with the student about this sentence and its meaning, we came up with an analogous sentence that expressed his intention:


            (55)  Housing might be the most important element (in our lives) because it gives us a place to sleep and enjoy ourselves.


Notice first that the original RC was meant to establish a semantic and circumstantial frame within which the main clause was to be interpreted.  This is very typical of Japanese and maybe of Korean too.  The next RC, taken from the same essay, is meant to be the thesis statement and comes at the end of the first paragraph of (54) above. 


            (54b)  The house which is influenced by the environment can be said to have a big difference between Korea and America.


Again, one is forced to interpret this RC as being nonrestrictive.  I read this sentence to several native speakers of English changing the subject to Housing and giving it the proper nonrestricitve intonation.  They all agreed it was better than the first, but still thought the use of the NRRC strange in this context.  The student and I rewrote it as:9


            (56)  The difference between housing in Korea and America is mainly due to the influence of the environment.


Both of these RCs have been used to convey very clear causal relations.  The question is why RCs like these were practically non-existent in the 66 Korean compositions chosen for this study. 

            The answer may lie in how the data was collected.  The compositions in this study were taken from the placement tests given at the English Language Institute.  This was done because of the large body of data available and the controlled nature in which it was collected.  These tests determine what level the student will be studying in for the next 3-4 months and so are usually taken quite seriously.  But, this fact may have impacted the data in a negative way with respect to RC production. 

            Park, Hyug-Bum used a number of RCs in every composition that he did in my class.  This student and I had frequently talked about improving his ability to use RCs and the different ways to use them.  He worked very hard at this even looking for opportunities to use RCs in his compositions.  But, he would frequently express the difficulty he had deciding when and how to use them.  On the midterm placement exam, after 8 weeks of instruction and much improvement in this area, Mr. Park did not use even one RC.  He seems to have avoided the construction because of the nature of the task, remembering his past difficulties.

            In summary, it appears that function is definitely an area of difficulty that Korean students face when acquiring the English RC.  While nothing in this study can invalidate the position which states that syntactic processing, the change from a Left Branching language to a Right Branching language, is responsible for the underproduction of the Korean students, we do have many reasons to believe that it is the use of the form and not the form itself which gives rise to a great deal of the avoidance phenomenon first witnessed by Schacter.     

Next: Findings Part 3