Introduction
Central to second language acquisition (SLA) theory, research, and pedagogy is the identification of the variety of input (language reception) and output (language production) that best leads second language learners toward mastery of their target language.  Researchers over the past 20 years have sought to corroborate the effects of input and output and many have produced evidence that naturalistic forms of input (such as immersion in an environment where the target language is spoken or a classroom that simulates an immersion experience) and output are not enough for second language learners to acquire accurate syntax.  Instead, many advocate that instructors provide input and opportunity for output that focuses learners’ attention on certain grammatical forms.  It is believed that a heightened awareness of such forms will ultimately lead to their acquisition.  However, this is an empirical issue yet to be resolved and this study provides evidence that improved written form is possible via naturalistic input and output.

Krashen’s (1985) Input Hypothesis provided a theoretical framework that claimed that comprehensible naturalistic input was a path toward second language acquisition.  Rapid changes followed with communicative methodologies and immersion programs replacing more behavioristic pedagogies (Harley, 1993).  Such programs were found to be deficient, however; learners became fluent and communicative but their syntax was still flawed (Hammerley, 1987; Harley, 1993).  Naturalistic input was viewed as incomplete; instead, input which focuses on form within communicative environments began to be explored with positive results (Rutherford & Sharwood Smith, 1985; Sharwood Smith, 1991, 1993; Long, 1991; VanPatten & Cadierno, 1993; VanPatten & Sanz, 1995; Doughty & Williams, 1998a; Ellis, R., 2001).

Swain also countered Krashen with the Output Hypothesis (1985, 1993, 1995) which maintains that comprehensible input is not enough; it is output that pushes learners into using language that will force them to recognize certain limitations in their language and rectify their faults either by serving: 1) a hypotheses-testing function about language via their interlocutors’ feedback; 2) a metalinguistic function which causes the learners to reflect on their language and process their syntax more thoroughly; and 3) a noticing or triggering function as learners are guided in to producing certain linguistic forms in order for them to “notice” where they have gone wrong.  Instructors guide learners into producing certain forms which will lead to their acquisition by “pushing” learners to notice their linguistic limitations.  As such, Swain also advocates focus on form and believes that learners can benefit significantly from focus on form within communicative classroom settings (1995).

While focus on form has been shown to produce short-term gains in the accurate production or recognition of certain linguistic structures, the long-term effects of such instruction have not been confirmed and evidence that such input is necessary is not conclusive (White, 1990; White, Lightbown, Spada, and Ranta, 1991; Carroll and Swain, 1993; Mackey et. al, 2004; Sheen, R., 2003).  Ron Sheen (2003) goes so far as to call focus on form “a myth in the making.”  Questions about the long-term effects of instruction that focuses on form beg further examination of a second theoretical position that argues that primary linguistic data, or naturalistic input, is all that language internal mechanisms need for acquisition to occur.  Therefore, learned linguistic knowledge via explicit focus on certain forms may result in short-term metalinguistic knowledge, but may not affect learners’ underlying linguistic competence, the system of Universal Grammar (UG), believed to be present in all humans at birth (Schwartz, 1993; Schwartz and Gubala-Ryzak, 1992; Birdsong, D., 2006). 

The goal of this paper is to revisit the concept of naturalistic input and output that does not seek to focus learners’ attention on any specific grammatical form by:  1) examining research that concludes that input and output that focuses on form are beneficial and exposing the flaws and limitations of these studies, and, 2) presenting a study that demonstrates enhanced mastery of target language verb forms in writing by using a methodology that has become known as “Fluency First” (MacGowan Gilhooley, 1991a).  This methodology utilizes naturalistic input and output in the form of extensive reading and writing and, while limited in scope, this study demonstrates significant gains in accurate use of verb forms by 19 student writers without any focus on form.
 

Input in SLA
Krashen’s influential Input Hypothesis (1981, 1985, 1989) suggested that comprehensible input that was just a notch beyond learners’ current competence would allow them to progress along a natural path toward long-term acquisition.  Pedagogies adjusted and immersion programs and communicative curricula were widely adopted.  However, Krashen came under attack (Gregg, 1984; Faerch and Kasper, 1986a; Sharwood Smith, 1985; White1987a; Gass, 1988; Ellis 1990a) mainly over his assertion that comprehending input and acquiring language are linked.  Many researchers saw second language acquisition as far more complex with factors such as age, motivation, first language literacy, attitude, and cultural assimilation as being also influential.  French immersion programs in Canada have led to fluency but not the mastery of certain grammatical forms (Harley, 1989a, 1993; Harley and Swain, 1984; Hammerley, 1987).   Despite passionate criticism of the attacks on French immersion programs (Collier, 1992), many researchers and teachers now consider it essential that students must have their attention focused on certain forms in order to make them aware of structures they may not be producing accurately.

 “Focus on form” instruction has come to mean drawing learners’ attention to a particular linguistic form while maintaining emphasis on meaning and use (Doughty and Williams, 1998; Ellis, 2002). I will, therefore, use the term focus on form throughout this paper to mean any attempt to draw learners’ attention to a specific linguistic form within the context of a meaning-driven communicative environment. 
 

Studies Investigating Input that Focuses on Form

Many authors of prior studies have used widely divergent methodologies to investigate the effects of input.  Most of these studies (Harley, 1989; Lightbown & Spada, 1990; Doughty, 1991; White et. al., 1991; White, 1992; Spada & Lightbown, 1993; VanPatten & Cadierno, 1993; Izumi & Lakshmanan, 1998) have three major methodological shortcomings:  1) they fail to gather longitudinal data; 2) they use grammar-related production tasks with little or no spontaneously produced language to document the effects of instructional input; 3) they rely on special methods; different instructors; and unusual tasks, equipment, schedules, and/or materials that draw students’ attention to structures being investigated.  Learners in these studies, I argue, may have been forced to focus on the explicit grammatical information being provided and, thus, may have performed better on assessment tasks investigating the effects of input, often administered hours or days after the instruction has been provided, than they would have if spontaneous, naturalistic language data had been collected.  

Such assessments investigating the effects of input are demonstrating the short-term effects of input that focuses on form but shed no light on whether the effects of such input are transient or truly acquired as part of the learner’s internalized system.  It appears that the effects of form-focused instruction may be successful only if certain structures are acquirable for individual learners (Dyson, 1996).  There is also extreme variation in the frequency and characteristics of form-focused instruction (Loewen, 2003; Mackey et.al., 2004) that may have both advantages and disadvantages for different learners (Ellis, et. al., 2002).   A review of form-focused instruction and how it affects the acquisition of implicit knowledge (Ellis, 2002) supports this claim by suggesting that form-focused instruction is influential if the choice of the structure and the extent of the instruction are considered as variables.  There certainly must be further study of how instructional conditions must be met for long-term alteration of learners’ linguistic knowledge to change. 

It is impossible within the scope of this paper to highlight the methods utilized in each of the studies cited above.  One example of a research method that is indicative of the problems that I have identified in many studies investigating the effects of input that focuses on form can be noted, however, in the Doughty (1991) study which investigates the effects of such instruction on the acquisition of English relative clauses by 20 international students studying at an intensive English institute in the U.S.  Improvement from pre-test to post-test in the ability to use relative clauses was compared using a control group which was exposed to relative clauses but received no instruction in relativization and an experimental group which was exposed to relative clauses and were also provided with additional instruction which focused on relativization in English. 

Learners were pre-tested and post-tested using a variety of focused written and oral tasks.  The written measures included two grammaticality judgment tasks, and two sentence combination tasks.  The oral measure included describing sets of similar pictures depicting people involved in some sort of action, and another picture depicting a park scene. 

Learners attended “ten working days” of treatment, one “lesson” per day.  Lessons were accessed via computer software and were completed by participants on an individual basis with the researcher present to answer any questions.  Learners were tested daily to ensure that they comprehended the lesson material.  At the end of the experimental instruction period, all participants were tested immediately using two of the original written test sections and two additional sections which merely substituted different lexical items.  The sections of the oral post-tests were those from the pre-test presented in different orders. 

Results showed the instructed group improving greatly in their use of English relative clauses and found control participants experiencing a significantly smaller gain than that observed in the experimental group. The improvement in the experimental group was attributed to the instructional techniques that provided form-focused instruction on relativization.

The claims that this study makes about form-focused instruction via participants’ interaction with instructional computer software are questionable.  While the input has been provided in regular intervals, it was provided over only a ten-day period.    Additionally, the instructional input is certainly drawing learners’ attention to the form being taught as the input is administered in a special place, via special equipment.  These special conditions, while drawing learners’ attention to the target structure, may be making them very aware of what is being presented and subsequently assessed.  As the duration of the study is ten days, the longitudinal effects of this input have not been ascertained.  Additionally, the assessment tasks do not ask the learners to produce any spontaneous or authentic language.   One or more of these methodological concerns can be seen in all of the studies cited above and should be considered when the benefits of focus on form instruction in SLA are touted.  Parallel concerns can be seen in studies investigating the effects of output.  

Output and SLA

Equally important and influential in terms of both theory and pedagogy is the issue of output.  Swain’s Output Hypothesis (1985, 1993, 1995) convincingly suggests that comprehensible input is not enough for learners to fully develop proficiency in their second language.  Learners must also be forced to produce “comprehensible output” for both fluency and accuracy in the target language to occur.  Where comprehension can occur without much syntactic analysis on the part of the learner, output forces the learner to produce syntactic forms, which allow communication to take place.  Furthermore, via output, learners have to recognize syntactic problems and remedy them (Swain, 1985).

Swain’s original Output Hypothesis has been refined (Swain, 1993, 1995) to include three functions of output.  The first, a hypothesis-testing function, claims that learners test their hypotheses about language by producing output and gauging the success of their utterances against the response of their interlocutors.  If their utterances are well formed, successful communication will transpire; if not, their interlocutors will signal that modifications in output must be made. 

The second function of output is a metalinguistic function that enables learners to become aware of their output and “control and internalize linguistic knowledge” (Swain, 1995, p. 126).  As such, learners have to process the linguistic information that they want to produce which may deepen their awareness of the form and rules of the target language.  This type of syntactic processing may lead to the type of modifications of output that facilitates acquisition (Swain, 1995). 

These first two functions are consistent with a model of output that relies on processes that allow internal language learning mechanisms to function.  These two functions can be seen as part of the process of language acquisition that both first and second language learners must utilize.  Indeed, both of these functions can be seen as operable without enhanced or explicit information on the part of an instructor.  Learners must negotiate meaning with interlocutors in order to communicate, and the negotiation and production of certain linguistic forms is certainly a part of any communicative process.  As such, these functions of output can be seen as vital to all language acquisition.

The third function, however, seeks to force output through derived exercises that force learners to notice certain forms.  This “noticing/triggering function” (Swain, 1995, 1998) leads learners to notice a gap between what they want to say and what they can say.  It is output, or production of language, that forces learners to notice the limitations in their language which would trigger them to modify their output in order to make it more syntactically accurate (Swain, 1995).  Production of the target language may be “the trigger that forces the learner to pay attention to the means of expression needed in order to successfully convey his or her own intended meaning” (Swain, 1995, p. 249). 

Swain has further stated (1993) that learners “need to reflect on their output and consider ways of modifying it to enhance comprehensibility, appropriateness and accuracy” (p. 161).  As such, Swain is an advocate of focus on form.  Studies investigating the effects of output on learners (Izumi et.al., 1999, Izumi and Bigelow, 2000) have tried to corroborate Swain’s claim with limited results and such studies can be questioned for many of the same methodological practices as those seen in studies examining the effects of input.

Studies Investigating the Effects of Output
Two studies (Izumi, Bigelow, Fujiwara, & Fearnow, 1999; Izumi and Bigelow, 2000) use the same experimental design to test the noticing function of output yet have not been able to produce conclusive evidence to support Swain’s hypothesis.  In both of these studies, the researchers rely on heavily contrived tasks that don’t fully succeed in drawing learners’ attention to the target form, in these cases the past hypothetical conditional.  Learners, again, are not provided with the opportunity to demonstrate that the structures in question have become part of their internalized systems as they have not been given the opportunity to produce the structure naturalistically.  Nor are the effects of the experimental treatments been examined beyond the month that each of these studies lasted. 

Instructional input in these studies has not been provided to learners through regular exposure to target language structures.  Izumi et. al. (1999) and Izumi and Bigelow (2000) present both their experimental and control groups “exposure to the input that contained many instances of the target form used in context” (p. 249) for a total of  5 and 4.5 total hours respectively.  While this can be seen as somewhat naturalistic input, the choice of vehicle to elicit the output of the target structure (the past hypothetical conditional) seems contrived.  Izumi and Bigelow (2000) are working within the context of a “process approach” to composition as this is the approach being used in these learners’ writing class.  Yet, it is highly unlikely that 80% of the sentences in a “model essay” would contain the past hypothetical conditional, even in the most focused assignment.  As such, this is a contrived assignment, and a poor example of effective writing; yet this forced type of output is being used to argue the benefits of focus on form.

Additionally, grammar-related as well as spontaneous production tasks are not used to assess the effects of both input and output.  Both of these studies use an essay writing task that is given to the experimental groups in both studies yet, as previously stated, this task is extremely contrived.  Additionally, the text reconstruction task that is also used is grammar-related and, like the essay-writing task, contains a high percentage of sentences containing the target structure that learners are asked to reconstruct as output.  This task is decontextualized and grammar-related as it supplies the learner with the structure to be manipulated and will, therefore be prompting learners’ use of explicit knowledge.  Birdsong (1989, 2006) points out that such grammar-related tasks are likely to be affected by differences in metalinguistic skills.  It seems logical that the astute adult learner will recognize this type of task as manipulative and will likely see it as foreign to what they would normally be expected to do in a process writing class.  As such, many will be using their metalinguistic skills to complete the task provided by the researcher.  Whether or not the participants in the Izumi et. al. and the Izumi and Bigelow studies are aware of what is being investigated and whether such awareness is affecting the performance of these learners must be called into question.

As seen in other studies investigating input in SLA, the effects of input on acquisition have not included longitudinal data.  The duration of the entirety of the Izumi and Bigelow study was 4.5 hours and the Izumi et. al. study 5 hours - each spread over one month.  It does not seem that this can be enough time to document whether long-term gains have been made. 

Also, the learners in the experimental group are distracted from their usual writing class by new instructors with different tasks that clearly focus on one structure.  In this case the past hypothetical conditional is being used in one task 80% of the time.  It seems safe to hypothesize that these learners will be prone to “notice” it more (one of the goals) but also will be using this metalinguistic awareness to their advantage in the short-term.  The effects of such awareness on the learners’ performance on experimental tasks seems an inefficient predictor of the long-term benefits of such “noticing.”  Until longitudinal data is collected and learners produce the target structure naturally after time, the effects of such output can only be seen as short-term and transient. 

Despite the inadequacies of Izumi and Bigelow’s research design, their results lead to another interesting point of discussion.  They found very little difference in the performance of their experimental and comparison groups, despite their use of “output” tasks for the experimental group (Izumi and Bigelow, 2000, p. 260 – 261).  This begs the question as to whether manipulated or enhanced input and output that focuses on form actually leads to long-term acquisition of these target structures or whether such gains are simply results of short-term cognitive strategies.

The in-vogue position regarding enhanced input, output, and focus on form as evidenced by many recent studies is that it must be beneficial.  But, in addition to the methodological flaws associated with many studies, another truth is often ignored:  even with instruction that focuses on form over long periods of time, many second language learners never achieve mastery of form in the target language.  This should encourage a review of whether naturalistic forms of input and output can lead to evidence of enhanced mastery of form. 

As 20 years of research has demonstrated, the effects of input and output on a learner’s developing linguistic system are complex and difficult to document.  Some of the flaws of prior studies have been highlighted above and should be taken into consideration when evaluating research design and results.  The purpose of this study, however, is simply to re-examine whether naturalistic forms of input and output in the form of extensive reading and writing can lead to increased syntactic accuracy in the use of verbs by college-age second language writers.   
 

METHOD

Participants

The 19 participants in this study were members of two sections of college-level English as a Second Language (ESL) writing classes at a four-year state college.  These matriculated students were placed via a holistically scored writing sample in a class that is entitled English as Second Language Writing I.  Students who take this class generally proceed to English as a Second Language Writing II which is the equivalent of English 101, or Composition I, a class taken by all of the college’s freshmen.  Each of the participants is between 18 and 22 years of age and has studied English formally from between one and seven years.  The groups were a mix of international students and students who permanently reside in the United States and the duration of their time in the U.S. also varied from three months to five years.  They were from Japan, Vietnam, The Dominican Republic, Ukraine, and Taiwan.

The “Fluency First” Method

The “Fluency First” teaching methodology used in these ESL writing classes was pioneered by at the City College of New York (CCNY) in the late 1980’s as an attempt to develop an approach to writing that emphasized fluency before clarity and correctness (MacGowan Gilhooly, 1991a).  This model is based on the idea that school programs that emphasize correctness before fluency are inhibiting young writers from using language before they even begin (Mayher, Lester, & Pradl, 1983).  A better way to teach writing, these authors reason, is to parallel the way children learn to speak naturally – by emphasizing meaning and communication over correctness.  They emphasize that fluency, clarity, and correctness are interconnected aspects of successful language use that writers cycle “over and over as they mature” (p. 54).

These ideas were adapted by CCNY professors MacGowan Gilhooly and Rorschach for their college ESL program as a response to what they saw as ESL instruction which gives “uniformly paced grammar instruction, exercises to practice that grammar, and intensive work on short, and unrelated reading and writing assignments with limited vocabulary exposure” (MacGowan Gilhooly, 1991a).   In response, the Fluency First program emphasizes that second language best develops in ways similar to first language and, therefore, needs similar types and quantities of language; and literacy in any language develops in ways that are similar to oral language development (Mlynarczyk, 1998). 

By adhering to these theoretical premises, the Fluency First classroom offers students extensive exposure to language in the form of reading and writing - naturalistic forms of input and output.  During a typical 15-week semester students read approximately 1,000 pages of popular literature, either fiction or non-fiction.  Additionally, they respond to their reading by completing daily “double-entry journals” where they copy a passage that piques their interest on the left-hand side of the page, and respond to that passage (or passages if necessary) in their own words on the right hand side of the page.  They complete a page a day, five to seven days per week, which results in between 75 and 100 pages of reading journal per semester.  Students are simultaneously working on five essays of approximately five pages, each taken through at least three drafts during the course of the semester.  They also freewrite in class for 10 minutes per day and may have other brief in-class writing tasks. 

Classroom communication is further enhanced by students working together in groups either discussing their reading journal responses and negotiating comprehension of the text or reading and responding to drafts of each other’s pieces of writing.  The remainder of time in class may be spent on teacher-led group discussion of the novel or presentation of strategies to improve, revise, or edit writing.  Students are encouraged to ask questions regarding specific grammatical points but rarely is grammar presented and there is no effort on the part of the instructor to focus on form.  During the editing or correctness stage of work on their individual pieces of writing, students are again encouraged to work on correctness but they rarely ask for help in this area during class time.

While there is little quantifiable data regarding the success of this model, most students and teachers at CCNY who have used this approach have been rewarded with dramatic improvement in students’ reading and writing.   The passing rate on the mandatory reading test of the two-thirds of all ESL students taught using Fluency First  rose by nearly 100% over several years;  the passing rate on the required 50-minute essay exam rose from around 32% to 69% over the same several year period (Mlynarczyk, 1998).  The percentage of students who passed freshman composition on their first attempt also rose from 57% to 86% during an eight-year period (Rorschach et. al., 1992). 

In keeping with the Fluency First model, the 19 students who participated in this study were in class for nine hours per week during the course of a 15-week semester.  The two sections of ESL Writing I were taught by two different instructors, one being the researcher.  However, both instructors collaborated on the amount of reading and writing assigned and the format of the class and assignments required were comparable.  All participants read approximately 900 pages of popular fiction during the semester and wrote approximately 75 pages of double-entry journal (in one section, students read The Green Mile by Stephen King, and Jurassic Park by Michael Crichton and in the other class, students read A Scream on the Water by Margaret Press, and The Prince of Tides by Pat Conroy).  Additionally, they did 10 minutes of freewriting at the start of every class.  At the end of the semester these students also presented five pieces of polished writing, between 25 and 35 pages, which had been taken through two or three drafts each.

Table 1 below places the “Fluency First” methodology in a broad historical context and general overview of language teaching.

Table 1 – A Broad Overview of Language Teaching
 

Examination of Verb Use

In order to best determine the effects of naturalistic input via extensive reading and writing using the Fluency First model, I examined participants’ use of verbs in their reading response journals.  I chose the reading journals to study as students wrote their responses to their reading on a daily basis and did not revise or edit these journal entries.  As they were using their reading journals as a starting-point for small-group discussions in class and they were evaluated on completeness of these journals, most of them were prepared with daily entries.  Students were asked to keep their journals and they handed them in as part of this study at the end of the semester. 
 

I chose errors in verb use to analyze, as correct verb use is critical to effective communication and central to the syntax of every sentence.  As such, verbs are easy to locate and identify and their large numbers best lead to the quantification of errors in usage.  The first 100 verbs in each participant’s journal (written in weeks one or two of the semester) were identified and coded as were the last 100 verbs of each journal (written in weeks 14 or 15 of the semester).  Verbs were coded as having either correct usage or as having one of eight patterns of error.  The types of error that were identified are: 

1.  inappropriate verb tense

Ex:  1a. I have the similar experience once.

        2a. (When I was a child) we can wear different clothes.

2.  missing subject

Ex:  2a. Secondly, canceled a dancing date.

3.  missing or extra auxiliary verb

Ex:  3a. In the US people are tend to change their jobs.

        3b. What Tom think about his sister’s feminism?

4.  inaccurate verb form or morphology

Ex:  4a. It could have happen to him.

        4b. The guy must not imposed on her feelings.

5.  conjugation of an infinitive

Ex:  5a.  People use to said one thing but they have something more in mind.

6.  missing copula or main verb

Ex:  6a. Something telling me that it just another day with lying Tom.
7. faulty agreement

Ex:  7a.  She have been there often.  

8. word order

Ex:  8a.  She have should done it.


Methodological Criteria

 As I have highlighted problems with prior studies that used controversial means of providing input and eliciting output, I have sought to:  1) use both authentic input and spontaneous output to analyze the language of learners in this study; 2) collect data longitudinally over the course of a 15-week academic semester and, 3) eliminate the use of special methods, different instructors, or unusual tasks, equipment, or materials.  I propose that these be minimum standards that researchers should adopt when investigating the effects of input and output on SLA.

RESULTS

Overall Reduction of verb-use errors

Participants in this study showed significant overall improvement in reducing their number of errors in verb use over the course of the 15-week semester.  This provides evidence that naturalistic input and output in the form of extensive reading and writing does result in increased syntactic accuracy.  Of the 19 participants, all but three showed improvement in their overall error rate in verb use as written in their double-entry reading response journals at the end of the semester.  Of 100 verbs counted and coded for verb use errors at the beginning of the semester, which I will call the pre-assessment, and at the end of the semester, which I will call the post-assessment, the participant with the highest rate of improvement made 28 fewer errors and had a 46% rate of improvement on the post-assessment.  The student with the lowest rate of improvement made only one fewer error on the post-assessment.  On the pre-assessment, participants averaged 27.6 out of 100 verb-use errors as compared to 19.8 out of 100 errors on the post-assessment.  The overall rate of improvement for all 19 participants was a rate of 8.4 errors fewer on the post-assessment, a 28% gain in reducing their verb use errors during the course of the study.  Of the three students that did not reduce their number of errors, one made four more errors on the post-assessment; one made one more error; and the third showed the same error rate at the end of the semester.  

Table 2 below provides the total number of errors on the pre and post-assessments for all participants and the reduction in the total number of verb use errors.

Table 2 Overall Error Reduction

Number of errors

Totals                   525                               376                                     149

Average                27.6                              19.8                                     8.4                                   28%

* Negative numbers in these two cases indicate an increase in errors made 

Figure 1 below provides the same data in graph form organized from left to right by participants with the highest to lowest number of errors at the beginning of the study.  While 16 of 19 participants showed a reduction in their number of errors, there does not appear to be a pattern of elimination of errors that correlates to higher number of errors on the pre-assessment. This discrepancy may have to do with initial level of English proficiency and overall ability to read and write.

Figure 1

Overall Error Reduction

In order to determine if the difference apparent in the data are statistically significant, a t-test was performed on participants’ overall error reduction.  This showed a significant error reduction of the naturalistic input and output via extensive reading and writing provided through the Fluency First methodology (t=4.827, p<=.001).
 

Most Frequent Errors
In examining the type of error that participants made most frequently, an even greater rate of success is shown.  Of the 19 participants, 17 showed an improved rate of grammatical accuracy in their use of verbs in the category that they made the most errors in the pre-assessment.  Of the categories of error listed above, the most prevalent in these participants are 1) inappropriate verb tense, the highest category for seven of the participants, and, 2) inaccurate verb form or morphology, the highest category for eleven of the participants.  Agreement was the highest category of error for one of the participants.  Table 3 below provides the number of errors in each participant’s highest category of error on the pre and post-assessments and the reduction in the number of these particular errors between the pre and post-assessments.  Overall, participants made an average of 10.7 errors of the same type in the pre-assessment, and 4.5 errors in the post-assessment, an average rate of improvement of 57% in reducing their most prevalent verb use errors during the course of the study.  Two of these learners showed complete elimination, or 100% improvement, in their highest category of error and four improved by 50% or more.

Table 3

Number of Errors in each Participant’s Highest Category

Number of Errors

Totals                                                204                                         85                                        119

Averages                                         10.7                                        4.5                                         6.3                               57%

* Negative numbers in this  case indicates an increase in errors made

Figure 2 provides the same information in graph form and highlights the improvement in verb use on the 19 participants in the category of error that they initially made most frequently.

Figure 2

Highest Category of Error

In order to determine if the difference apparent in the data are statistically significant, a t-test was performed on participants’ reduction of their highest category of error.  This showed a significant error reduction of the naturalistic input and output via extensive reading and writing provided through the Fluency First methodology (t=4.268, p<=.000). 

DISCUSSION

Overall Improvement
What participants in this study demonstrate is that improved written syntactic form, in this case in the use of verbs, is possible via naturalistic input and output in the form of extensive reading and writing.  Participants as a group show an average 28% improvement in eliminating all verb errors over the course of 15 weeks.  Why two of the participants made slightly fewer errors and one made the same number of overall errors on the pre-assessment than on the post-assessment can not be fully explained.  What we can assume is that there is variability in language proficiency, motivation, and personal and affective factors at play here as in all second language learning environments.  What is important to note is that improved form has occurred without any focus on form on the part of the instructor and without input or output that has been enhanced or manipulated in any way.  We also see participants using authentic language over four months, a period of time well beyond the average of most studies that claim positive effects of form-focused instruction or enhanced input and output.
 

Elimination of Specific Errors

Participants in this study made the most initial errors in use of inappropriate verb tense and in inaccurate verb form or morphology.  In close examination of these errors, patterns comparable to those seen in overall error rates emerge.  Of the 19 participants, all but two show improvement in their most prevalent errors over the course of 15 weeks, with several showing dramatic improvement.  Participants improved an average of 57% in eliminating their most frequent errors. 
 

What About More Problematic Structures?

This study has not looked at the improved syntactic accuracy in less-frequently heard and produced structures (such as the past hypothetical conditional) and, certainly, this is one of the arguments for the need for focus on form.  If these participants are only reading five examples of the past hypothetical conditional, for example (e.g., If it hadn’t rained, I would have gone swimming), over the course of 15 weeks, it is unlikely that, if they have not mastered that structure, they will be able to do so with such a limited amount of input and potentially no attempted output of this structure. 

What I do argue, however, is that prior studies, because of flawed and inconsistent methodology that does not include either longitudinal data or spontaneous and natural use of language, cannot adequately demonstrate whether enhanced forms of input and output have resulted in long-term alteration of the second language of the learner. 
 

CONCLUSIONS

While experimental in scope, this study provides some evidence that naturalistic input and output via extensive reading and writing can lead to increased syntactic accuracy in verb use.  Although further research in this area is necessary, some quantifiable evidence now exists that demonstrates that second language learners do not need enhanced input and output that focuses on form in order to show gains in syntactic accuracy.   What this study does not differentiate, however, is whether input and output provided to the learners in this study are working in tandem or whether input or output may be more influential.  That is a question for future study.  Additionally, it would be beneficial to be able to work with students using this teaching framework for longer than 15 weeks in order to determine whether longer exposure to the target language would result in complete elimination of errors of a certain variety as is seen in this study. 

Another important issue that must be addressed is whether linguistic forms that are rare in naturalistic input, and therefore likely to be rare or non-existent in naturalistic output, can be eliminated without focus on such forms.  This brings us to the issue of quantity. How much input and /or output is enough?  Participants in this study each read approximately 350,000 words and wrote over 30,000 words during the course of 15 weeks.  Each sentence that they read and wrote presumably resulted in the attempted use of a verb.  For future study, it will be helpful to refine the categories constituting the highest number of errors used in this study, namely errors in verb tense and verbal morphology.  By coding these specific errors, additional evidence about whether errors in less frequently utilized forms are reduced or eliminated will appear.  If all of the verb tenses and verbal morphemes were coded and counted separately, clearer patterns of improvement would certainly emerge. 

Certainly, performing a similar study with a comparison group receiving focus on form on a particular structure or structures would also be relevant and beneficial.  What I have advocated in this paper, however, is that for such a study to be valid, certain methodological criteria would have to be met.  Data would have to be collected longitudinally with assessments taking place at least several months after any instructional input that attempts to focus learners on a particular form.  This would be to ensure that long-term gains are being made.  Spontaneous production tasks, either spoken or written, would also have to be utilized to ensure that learners are relying on their linguistic systems and not some other cognitive strategies.  

A continued dialogue between researchers and more refined and effective studies are needed to answer these questions concerning the effects of various forms of input and output that are so vital to second language acquisition research and pedagogy.  Only when studies that incorporate learners’ spontaneous use of language over time are able to document the positive effects of enhanced input and output will evidence exist that long-term alteration of the second language system of the learner has occurred.  Until then, gains attributed to form-focused instruction will be suspect and questioned as to whether they are temporary or transient bits of learned linguistic knowledge.  This study does, however, provide limited quantifiable evidence that naturalistic language data does slowly but surely result in improved syntactic form in second language learners.

REFERENCES

Birdsong, D. (1989).  Metalinguistic performance and interlanguage competence.  NewYork:  Springer.
 

Birdsong, D.  (2006).  Dominance, proficiency, and second language grammatical processing.  Applied Linguistics, 27, 1 – 3. 
 

Carroll, S. and Swain, M.  (1993).  Explicit and implicit negative feedback:  an empirical study of the learning of linguistic generalizations.   Studies in Second Language Acquisition, 15, 357 - 386.

Collier, V.  (1992).  The Canadian bilingual immersion debate:  a synthesis of research findings.  Studies in Second Language Acquisition, 14, 87 – 97.

Doughty, C. and Williams, J. (Eds.) (1998a).  Focus on form in classroom second language acquisition.  New York:  Cambridge University Press.
 

Dyson, Bronwyn.  (1996).  The debate on form-focused instruction:  a teacher’sperspective.  Australian Review of Applied Linguistics, 19:2, 59-78. 
 

Ellis, Rod. (1990a).   Instructed second language acquisition.  Oxford: Blackwell.

Ellis, Rod.  (2002).  Does form-focused instruction affect the acquisition of implicit Knowledge?  A review of the research.    Studies in Second Language Acquisition, 24:2, 223 – 236.

Ellis, R., Basturkmen, H., and Loewen, S.  (2002).  Doing focus-on-form.  System, 30:4, 419 – 433.

Faerch, C. & Kasper, G.  (1986a).  The role of comprehension in second language acquisition.  Applied Linguistics, 7, 257 – 74.

Gass, S.  (1988).  Integrating research areas:  a framework for second language studies. Applied Linguistics, 9, 198- 217.
 

Gregg, K.  (1984).  Krashen’s monitor and Occam’s razor.  Applied Linguistics, 5, 79 – 100.
 

Hammerley, H.  (1987).  The immersion approach:  litmus test of second language acquisition though classroom

communication.  Modern Language Journal, 71, 395 – 401.
 

Harley, B. (1989a).  Functional grammar in French immersion:  a classroom experiment.  Applied Linguistics, 19,  331-359.
 

Harley, B. (1993).  Instructional strategies and SLA in early French immersion.  Studies in Second Language Acquisition, 15,

245 – 260.
 

Harley, B. & Swain, M. (1884).  The interlanguage of immersion students and its implications for second language teaching.  In

A. Davies, C. Criper, & A. Howatt (Eds.), Interlanguage (pp. 291-311).  Edinburgh, Scotland:  Edinburgh University Press.
 

Harley, B.  (1993).  Instructional strategies and SLA in early French immersion. Studies in Second Language Acquisition, 15,  245-259.
 

Higgs, T. and Clifford, R.  (1982).  The push towards communication.  In Higgs, (Ed.), Curriculum, Competence, and the Foreign Language Teacher. Skokie, Illinois:  National Textbook Company.

Izumi, S., & Bigelow, M. (2000).  Does output promote noticing and second language acquisition?  TESOL  Quarterly, 31, 2, 239-276.

Izumi, S., Bigelow, M., Fujiwara, M., & Fearnow, S.  (1999).  Testing the output hypothesis:  Effects of output on noticing and second language acquisition.  Studies in Second Language Acquisition, 21, 421-452.
 

Izumi, S. and Lakshmanan, U. (1998).  Learnability, negative evidence, and the L2 acquisition of the English passive.   Second Language Research, 14, 1, 62 –101.
 

Krashen, S. (1985).  The input hypothesis:  issues and implications.  New York: Longman.
 

Loewen, S.  (2003).  Variation in the frequency and characteristics of incidental form.  Language Teaching

Research, 7: 3, 315 – 345. 
 

Lightbown, P. and Spada, N.  (1990).  Focus-on-form and corrective feedback in communicative language teaching.  Studies in Second Language Acquisition, 12,  429-44.
 

Long, M. (1991).  Focus on form:  a design feature in language teaching methodology.  In de Bot et. al. (Eds.)  Foreign Language Research In Cross-Cultural Perspectives.  Amsterdam:  John Benjamins.
 

MacGowan Gilhooly, A. (1991a).  Fluency before correctness:  A whole language experiment in college ESL.  College ESL,
1, 37-47.
 

Mackey, A., Polio, C., and McDonough, K.  (2004).  The relationship between experience, education, and teacher’s use of incidental focus on form techniques.  Language Teaching Research, 8:3, 55 – 81. 
 

Mayher, J. S., Lester, N., Pradl, G. (1983).  Learning to write/writing to learn.  Portsmouth, NH:  Boynton Cook/Heinemann.
 

Mlynarczyk, R. W.  (1998).  Fluency First in the ESL classroom:  An integrated approach.  In Smoke, T. (Ed.) Adult ESL:  Power, pedagogy, and participation in classroom and community programs.  Mahwah NJ:  Erlbaum.
 

Rorschach, E., Tillyer, A., & Verdi, G.  (1992).   Research on ESL composition instruction:  the Fluency First approach.  Paper presented at the 26^th annual meeting of the Teachers of English to Speakers of Other Languages, Vancouver, British Columbia, Canada (ERIC Document Reproduction No. ED 350 848).
 

Rutherford, W. and Sharwood Smith, M.  (1985).  Consciousness raising and universal grammar.  Applied Linguistics,
6, 274- 82.
 

Schwartz, B. (1993).  On explicit and negative data effecting and affecting competence and linguistic behavior.  Studies in

Second Language Acquisition,15, 147-163.
 

Schwartz, B. and Gubala-Ryzak, M. (1992).  Learnability and reorganization in L2A:  against negative evidence causing the

unleashing of verb movement.  Second Language Research, 8.1, 1 - 38.
 

Sharwood Smith, M. (1993).  Input enhancement in instructed SLA.  Studies in Second Language Acquisition, 15, 165-179.
 

Sharwood Smith, M.  (1991).  Speaking to many minds:  on the relevance of different types of language information for the L2

learner.  Second Language Research, 7.2, 118-132.
 

Sharwood Smith, M.  (1985).  From input to intake: on argumentation in second language acquisition.  In S. Gass & C. Madden

(Eds.), Input in second language acquisition.  Rowley, MA:  Newbury House.
 

Sheen, Ron.  (2003).  Focus on form:  a myth in the making?  ELT Journal, 57:3, 225 – 233.
 

Swain, M. (1985).  Communicative competence:  some roles of comprehensible input and comprehensible output in its

development.  In S. Gass & C. Madden (Eds.), Input in second language acquisition (pp. 235-253).  Rowley, MA:  Newbury

House.
 

Swain, M. (1993).  The output hypothesis:  Just speaking and writing aren’t enough.  The Canadian Modern Language Review,

50, 158-164.  
 

Swain, M. (1995).  Three functions of output in second language learning.  In G. Cook & B. Seidlhofer (Eds.), Principles and

practice in applied linguistics:  Studies in honor of H. G. Widdowson (pp. 125-144).  Oxford:  Oxford University Press.
 

Swain, M.  (1998).  Focus on form through conscious reflection.  In C. Doughty &  J. Williams (Eds.), Focus on form in classroom

second language acquisition (pp. 64-81).  New York:  Cambridge University Press.
 

Swain, M., & Lapkin, S.  (1995).  Problems in output and the cognitive processes they generate:  A step toward second language

learning.  Applied Linguistics, 16, 371-391. 
 

Van Patten, B. & Cadierno, T.  (1993).  Explicit instruction and input processing.  Studies in Second Language Acquisition,
15, 225-243.
 

Van Patten, B. & Sanz, C. (1995).  From input to output:  processing instruction and communicative tasks. In F. Eckman, D. Highland, P. Lee, J. Mileham, & R. Rutowski Weber (Eds.), Second language acquisition:  Theory and pedagogy (pp. 169-185).  Mahwah, MJ:  Erlbaum.
 

White, L.  (1990).  Second language acquisition and universal grammar. Studies in Second Languag Acquisition, 12, 121-133. 
 

White, L.  (1991a).  Adverb placement in second language acquisition:  some effects of positive and  negative evidence in the

classroom.  Second Language Research, 7.2, 133 - 161.
 

White, L.  (1992b).  On triggering data in L2 acquisition:  a reply to Schwartz and Gubala-Ryzak.  Second Language Research,

8,  120 - 137.
 

White, L. (1987a).   Against comprehensible input:  the input hypothesis and the development of second language competence.  Applied Linguistics, 8, 95 – 110.
 

White, L., Spada, N., Lightbown, P., and Ranta, L. (1991).  Input enhancement and L2 question formation. Applied Linguistics, 12.4,  417-432.
 

Whitlow, C. J. (1997).  Assessing the effects of positive and negative input on second language acquisition: a study investigating the learnability of the English passive by speakers of Japanese. Unpublished Ph.D. dissertation, Boston University, Boston.

Whitlow, J. (2001).  Comments on Shinchi Izumi and Martha Bigelow’s “Does output promote noticing in second language acquisition?  Some methodological and theoretical considerations.”  TESOL Quarterly, 34, 2, 177 – 181.