Under review for:

Collentine, J.G. (Under review) "Development of Subjunctive and Complex-Syntactic Abilities among FL Spanish Learners." In B. Lafford and R. Salaberry (Eds.) Studies in Spanish Second Language Acquisition: the State of the Science. Washington, DC: Georgetown University Press.

This chapter critically reviews the research conducted to date on the acquisition of the subjunctive. While studies examining subjunctive development by native-speakers of Spanish and Spanish-English bilinguals in the US receive attention here, this chapter focuses its efforts on investigations involving foreign-language learners of Spanish. Because one of the key predictors of subjunctive development may be the learner’s abilities to generate complex utterances, this chapter also probes works on the acquisition of complex syntax and the effects of short-term memory limitations on the processing of such utterances. The author concludes by delineating some of the strengths and shortcomings of the studies reviewed, proposing suggestions for future research.

For many foreign-language (FL) learners of Spanish, one of the most unique grammatical constructs of the Spanish language is the subjunctive. The subjunctive is not highly productive in English, and so students have almost no L1 models with which to formulate hypotheses about its use in Spanish. This challenge is augmented by the subjunctive's linguistic complexities, denoting abstract concepts (e.g., marking irrealis events and states) and having a syntactic distribution that is largely limited subordinate clauses. The present review critically examines almost thirty years of research aimed at understanding the processes through which learners pass in the development of subjunctive abilities and the factors that interact with those processes.

While the primary purpose of this chapter is to provide an understanding of subjunctive acquisition in FL contexts, it contextualizes this foray with an analysis of important studies conducted on the acquisition of the subjunctive in first-language (L1) and bilingual contexts. Subsequently, the author examines the factors that affect subjunctive development in FL contexts as well as the variables influencing the development of abilities for processing complex syntax. Finally, the chapter outlines experimental design issues that limit the generalization of the conclusions that these L1, bilingual, and FL studies have offered.

Researchers have examined the effects of both internal and external factors on the acquisition of the subjunctive in L1 and bilingual contexts. The internal factors uncovered by these investigators involve the psycholinguistic mechanisms and maturational constraints that account for the developmental stages that children exhibit. The external factors relate to the effects of input (i.e., data) and sociolinguistic factors on the development of subjunctive abilities.

L1 Acquisition. Gili Gaya (1972) examines the acquisition of the subjunctive by preschool (i.e., three to five years) and school-age (i.e., five to ten years) children in Puerto Rico. Gili Gaya essentially posits that preschool children employ a lexical-cue strategy: that is, they use subjunctive forms with lexical phenomena that appear most frequently with the subjunctive (e.g., after querer que and para que). School-age children adopt a semantic strategy, generalizing the subjunctive to directives, volitives, and adverbial clauses with future time reference. Spanish speakers do not master mood selection until adolescence, when sociolinguistic factors (e.g., scholastic institutions) pressure them to conform to linguistic norms.

Blake (1983, 1985) examines the subjunctive development of Spanish natives ages 4 to 12 (N=134). His data confirm that school-age children largely employ a semantic strategy, limiting the subjunctive's function to "events that are not yet realized" (1985, p. 167), such as nominal clauses involving volition (e.g., Quiero que me hagas un favor), adjectival clauses (e.g., Quiero jugar con el equipo que sea mejor) and adverbial clauses entailing the future (e.g., Voy a hacer la tarea cuando vuelva a casa). Up to age 10 this semantic strategy competes with a lexical-cue strategy, which is especially robust in nominal clauses. Blake accounts for this with a cue-strength hypothesis of sorts. Specifically, adults demonstrate a great amount of variability in their use of the subjunctive in nominal clauses; consequently, children employ the subjunctive with those lexical phenomena that are reliably associated with the subjunctive in the input they receive.

Pérez-Leroux (1998) studies the effects of maturational factors on subjunctive development. Her data suggest that certain subjunctive functions (i.e., in adjectival clauses, e.g., La cocinera busca una gallina que ponga huevos) emerge only once a child can distinguish between their own beliefs and reality. Given that cognitive maturation mediates grammatical development, Pérez-Leroux concludes that children initially map deontic functions--whereas language reflects or affects changes within one's environment, such as in directives (cf. Palmer 1986)--onto the subjunctive. As they begin to distinguish between reality and belief, children assign it epistemic functions, which indicate the degree to which the child is committed to the truth value of a given proposition (e.g., No creo que sea cierto, Quiero una bola que sea grande; cf. Palmer 1986).

Bilingualism. Dialectologists studying US Spanish have provided insights into the factors that interact with subjunctive development in bilingual contexts (e.g., García and Terrell 1977; Hensey 1973, 1976; Sánchez 1972; Solé 1977). Floyd (1983) summarizes the results of these studies, essentially proposing a syntactic deficiency hypothesis to account for the underdeveloped mood-selection abilities of a group of Spanish/English bilinguals she studied in the early 1980s. She surmises that institutional prohibitions on the use of Spanish in public fora (e.g., in public schools) contributed to a delay in many bilinguals' syntactic development, which may, in turn, have hindered the development of subjunctive abilities.

Guitart (1982) tests an English-interference hypothesis: "The more a Spanish-English bilingual is influenced by English in his use of Spanish, the less he will use the subjunctive…[This hypothesis] is based on the fact that there is no mood contrast in such sentences in English." (p. 61). He compares the subjunctive abilities of three groups of bilingual adults (N=43), all of which differed in terms of their years of US residency. The data Guitart presents support his hypothesis, suggesting a negative correlation between residency in the US and subjunctive abilities.

Summary. This review indicates that both internal and external factors influence subjunctive development in L1 and bilingual contexts. Concerning internal factors, Pérez-Leroux's (1998) study indicates that the subjunctive's denotation--or range of connotations--is so abstract that acquisition is mediated by maturational constraints for a number of years. Still, even if a preschool child uses only a single semantic strategy when using the subjunctive, this strategy probably co-exists a lexical strategy throughout childhood (cf. Gili Gaya 1972; Blake 1983, 1985).

Regarding external factors, the subjunctive's distributional characteristics in the input that children receive appear to have important ramifications. Both Gili Gaya (1972) and Blake (1985) suggest that the strength of the association between the subjunctive and certain lexical items explain many developmental patterns. Yet, it is unclear whether cue frequency (i.e., the regularity with which children hear particular matrix-clause verbs in conjunction with the subjunctive) or cue strength (i.e., the reliability with which the subjunctive co-occurs with certain matrix-clause verbs) better accounts for behaviors observed.

Another important external factor is the strength of the institutional support for Spanish in the social contexts in which children develop (cf. Floyd 1983; Gili Gaya 1972; Guitart 1982). Studies indicate that, where there is institutional support, acquisition of the subjunctive nonetheless does not solidify until adolescence. This observation also implies that, apart from any semantic/pragmatic function, the subjunctive plays an important sociolinguistic role in the Spanish language.

Investigators studying internal factors affecting the acquisition of the subjunctive by FL learners have sought to account for developmental patterns within certain theories of L2 acquisition (e.g., the Monitor Model; cf. Krashen 1982). Those interested in external factors have not examined the effects of sociolinguistic factors on development as has been the case in L1 and bilingual studies; rather, they have sought to gauge the efficacy of instructional conditions and certain methodologies for promoting acquisition.

Internal Factors Affecting FL Subjunctive Development. Terrell et al. (1987) describe the subjunctive abilities of students completing one year of university-level studies (N=70) from the perspective of Krashen's (1982) Monitor Model. The learners' curriculum was a "cognitive approach" (Terrell et al. 1987, p. 21), whereas grammar explanations were followed by guided drills and practice as well as guided conversations. In a writing task, the subjects employed the subjunctive with 92% accuracy. In an oral (conversational) task, the learners' subjunctive accuracy averaged only 12.3%. It is also important to note that, in the oral task, the subjects produced few mood-selection contexts, or the pragmatic and syntactic conditions (i.e., subordinate clauses) that would necessitate the use of the subjunctive. Nonetheless, because the learners could only generate the subjunctive where they could monitor their performance (i.e., the writing task), Terrell et al. (1987) conclude that their subjects had "learned" but not "acquired" the subjunctive.

Stokes (1988) and Stokes and Krashen (1990) also attempt to account for subjunctive acquisition within the Monitor Model, wondering whether "acquired knowledge" can result from classroom instruction. The participants were advanced university-level students of Spanish, possessing varying amounts of formal Spanish study and time living in Spanish-speaking countries (N=27). For the study, the learners completed a sentence-completion task, which required them to repeat a matrix-clause (e.g., Mis padres quieren que yo…) and complete that clause with an original statement (e.g., Mis padres quieren que yo estudie italiano). The learners completed the task before and after "a week-long lesson on the subjunctive" (Stokes 1988, p. 706). (He avails no description of the materials or the methodological approach of the training component.) Stokes and Krashen (1990) examine the correlations between the participants' scores on the task, their number of semesters of classroom time, and their years of foreign residence. The analysis suggests that, while foreign residence correlates significantly with subjunctive abilities, classroom instruction does not. Stokes and Krashen found the same correlations even after the subjects completed the week-long subjunctive lesson. They conclude that "classroom instruction does not bring students far enough along to acquire the subjunctive." (Stokes and Krashen 1990, p. 806)

Collentine (1995) attempts to describe and account for the subjunctive abilities of learners completing the second year (i.e., the intermediate level) of university-level Spanish FL instruction (N=78) within Givón's (1979) Discourse Hypothesis. The learners' curriculum was proficiency-oriented (cf. Whitley 1993). Analyzing two oral-production tasks (i.e., one involving spontaneous speech and the other planned speech), Collentine suggests that most learners completing the intermediate level evince behaviors typical of the "presyntactic stage" of development, employing morphology unreliably and exhibiting poor subjunctive abilities even when they have a generous amount of time to produce utterances. Collentine employs a syntactic-deficiency hypothesis to explain the difficulties that these learners face with learning the subjunctive. Specifically, he posits that these learners are not yet at a point where they can process in short-term memory both complex syntax and the semantic/pragmatic relationships (e.g., volition, doubt/denial) existing between two clauses.

External Factors Affecting FL Subjunctive Development. Most studies examining external factors involved in subjunctive acquisition in FL contexts assume that meaningful input provides learners with the data necessary to build a grammatical system (Larsen-Freeman and Long 1991; Lee and VanPatten 1995; Schmidt 1990; VanPatten 1993, 1997). To be more specific, these studies adopt the view that the subset of the input that becomes intake directly feeds data to the developing grammatical system (cf. Leow 1993). Features of a grammatical phenomenon can become intake only if a learner "notices" (Schmidt 1990) or "detects" (Tomlin and Villa 1994) those features. Accordingly, investigators have examined from a variety of perspectives what shall be termed here the subjunctive intake disposition, which refers to the likelihood that elementary- and intermediate-level FL learners of Spanish will detect and/or intake subjunctive forms. Additionally, researchers have tested the efficacy of certain instructional techniques and methodologies for promoting subjunctive intake.

Before examining the subjunctive's intake disposition, it is important to outline the behaviors that are characteristic of detection and intake. Following VanPatten (1990), grammatical constructs place varying demands on one's short-term memory and attentional capacity, both of which are limited in capacity at any given moment. If learners detect exemplars of a grammatical structure for which they have little background knowledge, they have fewer processing resources to dedicate to the comprehension of the meaning of the sentence/passage surrounding those exemplars. Thus, if a new grammatical structure is associated with a decrease in comprehension, detection has probably occurred. The most common instrument for inferring the occurrence of intake has been the so-called recognition test: after having read or heard a passage containing exemplars of a targeted structure, learners indicate somehow whether they recall having encountered those exemplars.

Lee (1987) presents the first evidence that the subjunctive will tend to elude detection. The primary intent of his study is to assess the tenability of the premise that learners need early exposure to the subjunctive before they can comprehend reading passages containing forms such as tenga and compre. Two groups of FL learners of Spanish read a passage containing numerous subjunctive forms: a first-semester group (n=90) without prior subjunctive instruction and a second-semester group (n=90) with prior instruction. Lee (1987) finds that prior exposure to the subjunctive did not predict comprehension, as both groups understood the reading's content equally well. The learners without prior subjunctive instruction probably would have experienced less overall comprehension if they had detected the subjunctive.

Lee and Rodríguez (1997) provide insights into the effects of various factors on learners' intake of the subjunctive. The focus of their study is broader than simply examining subjunctive intake, as they investigate the interactions between three grammatical variables and reading comprehension: (1) mood, exposing intermediate-level learners of Spanish (N=120) to a passage with either proper uses of the subjunctive or with (erroneous) indicative forms (e.g., Es importante que se *prepara/prepare); (2) subordination, whereas subjects read a passage containing either subordinate or non-finite clauses (e.g., Es necesario que lo hagan/*hacerlo); and (3) vocabulary, whereas the learners read subjunctive and/or subordinate clauses with either known vocabulary or nonce terms. Lee and Rodríguez present data from both comprehension and recognition tests. Interestingly, passages containing subjunctive forms and/or complex syntax were no more difficult to comprehend than those containing simpler morphology and syntax. Nonetheless, subjunctive forms became intake less frequently than indicative forms, implying that the morphological differences between the present indicative and the present subjunctive are not perceptually salient to learners at the intermediate level. Interestingly, the learners were more likely to intake the subjunctive and subordinate-clause syntax when the vocabulary items encoding or associated with such grammatical information were familiar.

Leow (1993) provides another window into the external factors affecting subjunctive acquisition in a study in which he asks whether simplified input (i.e., a simulated-authentic passage as opposed to an authentic one) facilitates the intake of verbal morphology (i.e., the present perfect and the present subjunctive) at the first and second years of university study. Two groups of first-semester learners of Spanish (n=49) and two fourth-semester groups (n=88) read either a simplified or an authentic passage containing both constructs. The fourth-semester groups had previous exposure to both structures. Leow operationalized intake with a pretest/posttest recognition test. Simplified input did not facilitate intake at either level of proficiency, although the fourth-semester learners experienced significantly more intake of the targeted structures. Leow conjectures that advanced learners intake more grammatical information because they can dedicate more attentional resources to such phenomena than less advanced learners, who are primarily concerned with processing input for lexical content (i.e., for meaning). This suggests that learners must establish a certain linguistic foundation before they will intake morphological paradigms like the subjunctive. Leow (1995) replicates his 1993 study in the aural mode. Again, proficiency level rather than simplification accounted for intake. However, Leow (1995) reveals that FL learners of Spanish are less likely to intake the subjunctive than they are the present perfect in aural passages. Leow essentially argues that the phonological properties of the subjunctive are less salient to learners than the features of the present perfect.

Following Leow (1995), Collentine (1997) asks whether learners are more likely to detect so-called irregular subjunctive forms (e.g., sea, sepa, tenga, vaya) than regular subjunctive forms (e.g., compare, beba, viva). As verbal stimuli go, the former are perceptually more "novel" (cf. Cowan 1995). Intermediate-level learners of Spanish (N=30) utilized a computer application that measured whether irregular forms required more time to process and whether they interfered with comprehension, behaviors that VanPatten (1990) views as indications that intake has occurred. Collentine observed both behaviors, indicating that irregular forms do, in fact, increase the likelihood of subjunctive detection.

Research examining the efficacy of instructional techniques on subjunctive learning has primarily investigated whether "structured input" (cf. VanPatten 1993; VanPatten and Cadierno 1993) promotes acquisition. Structured-input tasks expose learners to sentences containing exemplars of a targeted structure. Intake of the targeted structure is likely to occur if the task is designed in such a way that the proper interpretation of a sentence's message largely depends on properly interpreting the connotation of that structure. For instance, in a lesson on indirect-object pronouns, a student might somehow indicate whether first-person singular (i.e., yo, the person talking) or second-person singular (i.e., tú, the person listening) is the agent of a sentence such as Te doy mi suéter favorito.

Collentine (1997) predicts that structured input cannot promote subjunctive intake in nominal clauses. From a logical standpoint, he argues, the modal information that the subjunctive connotes in such sentences is invariably encoded lexically into those sentences' main-clause verbs (e.g., Quiero que me hagas un favor, Dudo que tengan suficiente). However, a consideration of Pereira's (1996) work implies that Collentine's prediction is erroneous. Third-semester FL learners of Spanish (N=68) participated in Pereira's study, which--among other things--assessed the potential of structured input to promote subjunctive development in nominal clauses. The learners had no prior subjunctive instruction. A pretest, a posttest, and a delayed posttest entailing grammaticality judgements and a dialogue-completion exercise confirmed the efficacy of structured input in promoting subjunctive knowledge.

Woodson (1997) compares the efficacy of Processing Instruction (PI) (cf. VanPatten 1993)--a methodology that employs structured input--to what she terms Interactive Processing Instruction (IPI) for promoting subjunctive acquisition. IPI incorporates principles of the Interactionist Hypothesis (cf. Pica and Doughty 1985; Gass and Varonis 1994) and cooperative learning (cf. Pica et al. 1993). IPI prompts learners to focus on a structure's form and meaning in tasks where they share their interpretations of aural and written input with other learners. The crucial difference between PI and IPI, however, is that IPI encourages the production of a targeted structure; learners problem-solve in the target language (e.g., in jigsaw activities) in tasks where they must write or generate orally the targeted phenomenon. High school-level FL learners of Spanish (N=48) without prior subjunctive instruction participated in the study. Assessment entailed two written and two oral tasks. Woodson found PI and IPI to be equally effective.

Collentine (1998) compares the efficacy of PI and an output-oriented approach to grammar instruction on subjunctive development in adjectival clauses. The output-oriented instruction involved learners in meaningful speaking and writing tasks where they were prompted to produce the subjunctive. Second-semester FL learners of Spanish without prior subjunctive instruction participated in the study (N=54). Utilizing a pretest/posttest paradigm, an interpretation task (where learners identified whether particular uses of the subjunctive were appropriate given contextual clues) and a written production task assessed the treatments' effects. As Woodson (1997) found, the analysis indicated that the input-oriented PI and the output-oriented approaches are equally effective at promoting subjunctive development.

Summary. The research investigating internal factors (cf. Collentine, 1995; Terrell et al. 1987; Stokes 1988; Stokes and Krashen 1990) indicates that many classroom contexts do not promote the sort of automatized (or proceduralized) subjunctive knowledge that learners will need to make this structure a functional part of their IL until quite late in their development (cf. R. Ellis 1994).

Investigations into the external cognitive factors affecting subjunctive development suggest that, by and large, subjunctive forms elude detection and intake during the initial stages of acquisition. Perhaps the most important explanation resides in the phonology of the subjunctive. Many present-subjunctive forms do not differ significantly from their present indicative counterparts, possessing the same stems (e.g., canto versus cante) and inflections that learners already associate with the (present) indicative (e.g., bebe [indicative] versus cante [subjunctive]). Furthermore, the ability to interpret or generate the abstract semantic nuances that the subjunctive relates is probably not critical to most learners up through the intermediate level of instruction, who are primarily concerned with the core (i.e., propositional) content of utterances.

The research reviewed above reveals important factors whose consideration may yield more effective tasks and materials. First, Lee and Rodríguez (1997) as well as Collentine (1997) suggest that tasks promoting subjunctive intake must consider the type of subjunctive forms that they will include. So-called high frequency verbs (i.e., the most common verbs, which also tend to be the most irregular) possess some feature (either semantic or phonological) that partially offsets any inherent elusiveness. Second, while input-oriented techniques, such as those involving structured input, may be important facilitators in promoting subjunctive development (cf. VanPatten 1993), they may not be sufficient (cf. Pica and Doughty 1985; Gass and Varonis 1994; see also Swain 1985). While VanPatten (1993) asserts that production-oriented tasks constitute a means to reinforce rather than to affect grammatical development, the subjunctive may require production (i.e., speaking, writing) tasks (cf. Collentine 1998; Woodson 1997).

Finally, Leow's work (1995) suggests that subjunctive intake is mediated by the learner's overall proficiency, perhaps indicating that a certain linguistic foundation must be in place before subjunctive intake will occur reliably. Interestingly, Collentine (1995), Pereira (1996), as well as Terrell et al. (1987) conjecture that subjunctive instruction will only be beneficial once a learner can process complex syntax. Indeed, recall that Floyd (1983) explains the poor subjunctive abilities of the bilinguals in her study by essentially implying that their inabilities to generate complex syntax interacted with the acquisition of their subjunctive abilities. The next section explores the tenability of such a syntactic-foundation hypothesis.

The following reviews the tenability of the syntactic-foundation hypothesis, namely, that the acquisition of the ability to process complex syntax establishes certain prerequisite developmental conditions for the acquisition of the subjunctive. Unfortunately, the literature contains no direct support for this position, as no one to date has examined the relationship between complex syntactic abilities and the acquisition of subjunctive abilities in a FL context (or any other context). Still, in general, the following will show that syntactic abilities seem to outpace morphological abilities during IL development, although it is unclear whether this pattern results largely from linguistic or cognitive mechanisms.

The 1970s and 1980s saw a plethora of studies detailing the order in which learners acquire various grammatical entities, including morphology and syntax. R. Ellis (1987) summarized the findings of these studies, surmising that learners generally progress through four sequential stages: (1) the development of basic syntactic knowledge, such as SVO in English; (2) the acquisition of variant word order, such as knowing that subject-verb inversion equates to question formation; (3) the development of morphological knowledge; and, (4) the acquisition of knowledge relating to complex sentence structure.

Bardovi-Harlig and Bofman (1989) specifically studied the relationship between the acquisition of interlanguage (IL) morphology and syntax. They compared the syntactic and morphological errors of advanced learners of English, concluding that these learners committed fewer syntactic than morphological errors. Bardovi-Harlig and Bofman (1989) reasoned that, when the L1 and the L2 are Indo-European languages such as English, German, and Spanish, syntax develops before morphology because syntax enjoys greater crosslinguistic "stability" than morphology. That is, there are greater syntactic similarities between Indo-European languages than morphological similarities, which may give the acquisition of syntax an advantage.

The theory of Universal Grammar--UG--(cf. Cook 1994)-- predicts, in no uncertain terms, that syntactic knowledge will outpace morphological knowledge in some aspects of IL development. Firstly, syntactic knowledge, it is hypothesized, resides at the core of UG whereas morphological knowledge resides at the periphery. Thus, if UG is accessible to the adult acquiring a FL, he or she possesses a more powerful set of developmental principles for syntax than for morphology. Secondly, a central tenet of UG is the structure dependency principle: one does not combine morphemes (and so words) with great freedom; instead, their usage is limited to particular syntactic environments (e.g., campo cannot serve as a verb). Accordingly, if a learner has not developed the syntactic knowledge that surrounds the use of a given morpheme (e.g., gender agreement often depends on an understanding of how one forms a noun phrase in the target language), the learner may attend to it marginally in input or he or she may not contemplate using that morpheme in production.

SLA research informed by UG predicts that one's IL grammar will not generate verbal morphology until it can generate certain phrase-structure constituents. The Gradual Development Hypothesis (cf. Vainikka and Young-Scholten 1994, 1996) posits that learners do not initially generate functional projections such as IP—i.e., a constituent that carries information such as person, tense, and mood. Consequently, one cannot expect a learner's competence to intentionally produce certain morphology (i.e., which is not the case when one generates a formulaic chunk; Me gusta estudiar español) until the IL starts to encode utterances with the requisite syntactic structure.

Skiba and Dittmar (1992) examine the grammaticalization of syntactic and morphological knowledge in L2 learners of German. Skiba and Dittmar observe that learners' sensitivity to certain morphological features follows an important advancement in their syntactic abilities, namely, after they abandon the basic TOPIC-COMMENT word order (e.g., Juan…listo y simpático) in favor of a SUBJECT-PREDICATE strategy (e.g., Juan es listo y simpático). These researchers document that it is only once learners enter the SUBJECT-PREDICATE stage that they "try to modify verb forms morphologically" (p. 339). Skiba and Dittmar posit that morphological systems develop out of the assignment of syntactic features to individual lexical items, such as the assignment of [+N,-V] to terms such as casa and [-N,+V] to trabaja.

Nevertheless, cognitive perspectives on SLA suggest that syntax need not necessarily outpace morphology during IL development. Connectionism--a theory of learning with roots in the field of cognitive psychology--posits that generalized principles of learning rather than innate, language-specific principles (e.g., UG) account for acquisition (cf. N. Ellis 1999). The connectionist perspective makes two principal predictions about SLA. First, as Bardovi-Harlig and Bofman (1989) assert, the learner uses his or her L1 as a model for comprehending and forming L2/FL sentences (MacWhinney 1996). Second, where the L1 does not provide good models, the learner's working and phonological memory stores play a central role in processing utterances and building an L2 grammatical system (N. Ellis 1996).

Initially, for the native English speaker, since Spanish and English share many syntactic and few morphological features, syntactic development in Spanish will be accelerated whereas morphological development will be slow. However, the learner will eventually be confronted with evidence that English syntactic frames are not entirely reliable models for Spanish, and so some reorganization will occur (N. Ellis 1996). At this point, the connectionist perspective appears to predict, in contradistinction to UG, that knowledge of target-language morphology establishes conditions for the acquisition of complex syntactic knowledge (e.g., knowledge of how to form a dependent clause in the target language), since determining that a given string represents a syntactic frame depends on one's ability to parse the string's lexical and morphological features:

Connectionist accounts of SLA predict that certain types of short-term memory (STM) play especially critical roles in the development of knowledge of syntactic and non-local morphological structures. An important mechanism in the development of both syntactic and verbal morphological knowledge is short-term phonological memory (PSTM). Learners analyze and compare PSTM chunks to chunks stored in long-term memory (LTM) in order to extrapolate generalizations about the target language's grammatical system (N. Ellis 1996; MacWhinney 1982). One's ability to maintain and operate on information in working memory is also crucial for the acquisition of verbal morphology (King and Just 1991), as the referents for person/number morphemes are often only retrievable upon an examination of a given discourse's so-called thematic nodes, or slots in episodic memory that represent the actors/objects in a discourse (Givón 1990). These STM stores, whose capacity increases as the chunks of information that one can store there increase in size, is particularly important to consider in the acquisition of Spanish as a FL, as Spanish verbal morphology places "heavy demands on working memory and phonological rehearsal" (MacWhinney 1996, pp. 311-312).

Collentine (submitted) is the only work to date investigating whether there is a positive, linear relationship between the syntactic and morphological abilities of FL learners of Spanish. Advanced-level learners at least a semester beyond the intermediate level (N=30) participated in a study gauging their abilities to utilize syntactic and morphological cues to interpret aural passages. While the analysis indicated that increasing syntactic abilities correlate positively with increasing morphological abilities, the analysis did not confirm that the learners' syntactic abilities were superior to their morphological abilities. Furthermore, the most robust predictor of both syntactic and morphological behavior was the extent to which a learner could process long-distance dependencies (i.e., referential morphology, such as person and number, and inter-clausal syntactic relationships, such as subordination). Collentine conjectures that, while enhanced syntactic abilities may support the development of mood-selection abilities, researchers may need to consider whether the principal factor that impedes subjunctive development is neither syntactic nor morphological in nature; instead, both abilities may improve in tandem as learners' general abilities to process long-distance dependencies in short-term and working memory improve (cf. N. Ellis and Schmidt 1997).

Summary. Although some researchers examining the factors affecting the acquisition of the subjunctive surmise that the ability to process complex syntax establishes conditions that facilitate the development of subjunctive abilities (Collentine 1995; Pereira 1996; Terrell et al. 1987), no study has specifically studied this relationship. Still, it is noteworthy that the SLA literature documents that syntactic abilities tend to outpace morphological abilities during IL development, at least when the L1 and the L2/FL are Indo-European (Bardovi-Harlig and Bofman 1989; R. Ellis 1987). UG theory and its SLA students argue unequivocally that some syntactic knowledge is a prerequisite for the processing, and so development of, verbal morphology (Skiba and Dittmar 1992; Vainikka and Young-Scholten 1994, 1996). Yet, from the onset of acquisition, learners cannot process large phonetic chunks and long-distance dependencies in short-term memory, which may interact with one's capacity for processing both complex syntax and morphological structures such as the subjunctive (N. Ellis 1996; MacWhinney 1996). (Recall that the subjunctive often redundant the mark's pragmatic information that is "recoverable" in the main clause of a sentence in which it appears, Queremos que nos hagas un favor.) Of course, support for this explanation is essentially theoretical, with only one study--(Collentine, submitted)--providing empirical considerations.

Any comprehensive review of a body of research necessitates a consideration of the experimental design and the analytical tools supporting that research's findings. This section delineates general issues that future researchers must consider in interpreting the studies reviewed here and upon undertaking future studies. It is, however, essential to bear in mind that the field of applied linguistics has shifted in the past 30 years from favoring ethnographic approaches to research design and analysis to favoring decidedly experimental protocols. Thus, it is not entirely fair to expect that the tight controls placed on experiments today should have been applied to studies when the field was in its infancy.

The legitimacy of the conclusions extrapolated from any experiment is invariably mediated by the extent to which investigators control for variables and how they operationalize research questions. Researchers have tended to treat the subjunctive as a monolithic phenomenon (Blake, personal communication). This approach undoubtedly masks a range of developmental patterns that might be uncovered if, for instance, investigators were to compare developmental patterns of the subjunctive in nominal, adjectival, and adverbial clauses.

Investigators have rarely controlled for the amount of subjunctive knowledge that learners possessed prior to experiments testing the efficacy of certain methodologies to promote subjunctive development. For instance, Leow's (1993) two groups did not only differ in terms of proficiency level: his fourth-semester learners had prior exposure to the target structure, and so it is just as likely that their knowledge of the subjunctive also contributed to their intake of more subjunctive forms. Nor do Pereira (1996) or Woodson (1997) attempt to gauge their subjects' subjunctive abilities going into their respective treatments. Consequently, extrapolations from these studies must progress cautiously. Collentine (1998) is the only experiment to use pre-treatment subjunctive abilities as a co-variate in the analysis of the efficacy of a methodology.

Controlling for participants' prior knowledge of grammatical phenomena such as the subjunctive is especially important given that subjunctive development may interact with the development of one's overall grammatical abilities. Thus, researchers should minimally measure learners' grammatical abilities in some holistic fashion before administering any treatment and factor in those results as co-variates in their statistical analyses.

The legitimacy of any conclusion also depends on how investigators operationalize their research questions. First, the studies investigating the internal factors that account for subjunctive development routinely treat instruction as a monolithic phenomenon (cf. Collentine 1995; Terrell et al. 1987; Stokes 1988; Stokes and Krashen 1990). By all accounts, these studies sampled student populations that learned Spanish in curricula where form-focused instructional strategies predominated and so their conclusions potentially misrepresent the effects of meaning-focused curricula, such as those that employ the Natural Approach or Processing Instruction. Second, assessment-task measures are often problematic, since there is no consensus about which testing format(s) assess the current status of a learner's subjunctive knowledge with the greatest reliability and validity. The most notable design flaw in the studies reviewed here is found in Stokes (1988) and Stokes and Krashen (1990), who attempt to assess learners' acquired knowledge for mood selection with a sentence-completion task. Such a task is not naturalistic, and so it is unclear how their subjects' performance exhibits the types of cognitive processes and behaviors manifested in spontaneous language use (e.g., conversational tasks). Third, the research on L1 subjunctive development purports to account for the stages children experience (e.g., Blake 1983, 1985; Gili Gaya 1972; Pérez-Leroux 1998). Yet, the design of these studies was not longitudinal but rather cross-sectional, which paints an over-simplistic picture of one's developmental status at any given point (R. Ellis 1987).

The appropriateness of any study's statistical analyses is a crucial consideration upon contemplating the legitimacy of one's conclusions. Such analytical rigor is absent in the research conducted on subjunctive development in L1 and bilingual contexts as well as in much of the research on the internal factors that promote subjunctive development in a FL context. (See Stokes (1988) and Stokes and Krashen (1990) for notable exceptions.) Studies such as Collentine (1995) claim certain significant differences between FL learners' indicative and subjunctive abilities; yet, Collentine (1995) employs only non-parametric analyses, thus raising concerns about the generalizability of his conclusions to the student population as a whole. The best analytical models for future investigations are present in the research on the external variables accounting for FL subjunctive development (e.g., Collentine 1998; Lee and Rodríguez 1997; Leow 1993, 1995; Pereira, 1996) as well as in the analyses employed by Pérez-Leroux (1998).

Finally, it is important to keep in mind that hypotheses whose support is largely theoretical are tentative at best. Researchers such as Collentine (1995), Pereira (1996) and Terrell et al. (1987) have hypothesized that as one abilities to process complex syntax increase, his or her acquisition of the subjunctive will progress. However, no empirical evidence supports such assertions and indirect support is largely theoretical. Indeed, a consideration of a connectionist perspective on SLA makes converse predictions, namely, that the acquisition of certain morphological abilities establish the developmental conditions for certain syntactic abilities.

Perhaps the most important finding stemming from the three decades of subjunctive research is that learners do not acquire skills and knowledge for this construct in isolation of other aspects of their IL development, such as their syntactic development and their abilities to process long-distance dependencies in short-term memory. Furthermore, we are only beginning to understand the effects of instruction on subjunctive development. Perhaps the struggles that classroom learners have experienced to date are attributable to poorly designed instructional techniques and to a propensity of Spanish educators to promote subjunctive development prematurely. Researchers should nonetheless be cautious about extolling the virtues of recently developed instructional techniques, which tend to advocate either a largely input or a largely output-oriented approach to instruction (cf. VanPatten 1993). Many subjunctive forms appear to elude intake. In addition, evidence exists that production-oriented learning tasks are equally effective at promoting subjunctive development with equal effectiveness as input-oriented tasks. Finally, if subjunctive development interacts with various developmental processes, researchers will ultimately need to tease out how the array of developmental factors interact with different teaching techniques.

Of course, it is possible to argue that investigating subjunctive development focuses an extraordinary amount of resources on the acquisition of one grammatical phenomenon at the expense of studying phenomena whose developmental patterns are more generalizable (e.g., the acquisition of one's overall inflectional abilities). Yet, as the above review indicates, subjunctive research provides insight into the complex nature of L2 development. That said, it is not unreasonable to expect that this very line of research provides a different sort of generalized view of Spanish FL acquisition.

Bardovi-Harlig, K., and Bofman, T. 1989. Attainment of syntactic and morphological accuracy

Blake, R. 1983. Mood selection among Spanish speaking children, ages 4 to 12. The Bilingual

---. 1985. From research to the classroom. Hispania 68: 166-173.

Collentine, J. 1995. The development of complex syntax and mood-selection abilities by

---. 1997. Irregular verbs and noticing the Spanish subjunctive. Spanish Applied Linguistics 1:

---. 1998. Processing Instruction and the subjunctive. Hispania 81: 576-587.

---. Submitted. The Relationship between Syntactic and Morphological Abilities in FL Learners of

Cook, Vivian. 1994. Universal Grammar and the learning and teaching of second languages.

Cowan, N. 1995. Attention and Memory: An Integrated Framework. New York: Oxford

Ellis, Nick. 1996. Sequencing in SLA: Phonological memory, chunking, and points of order.

---. 1999. Cognitive approaches to SLA. Annual Review of Applied Linguistics 19: 22-42.

Ellis, N., and Schmidt, R. 1997. Morphology and longer-distance dependencies: laboratory

Ellis, R. 1987. Second language acquisition in context. Englewood Cliffs: Prentice Hall.

---. 1994. A theory of instructed second language acquisition. In Implicit and explicit learning

Floyd, M. 1983. Language acquisition and use of the subjunctive in Southwest Spanish. In

Fodor, J. D. 1983. The modularity of the mind: An essay on faculty psychology. Cambridge,

García, M., and Terrell, T. 1977. Is the mood in Spanish subject to variable constraints? In

Gass, S., and Varonis, E. 1994. Input, interaction, and second language production. Studies in

Gili Gaya, S. 1972. Estudios de lenguaje infantil. Barcelona: Bibliograf.

Guitart, J. 1982. On the use of the Spanish subjunctive among Spanish-English bilinguals.

Givón, T. 1979. On understanding grammar. New York: Academic Press.

---. 1990. Syntax: A functional-typological introduction: Volume II. Philadelphia: John

Hensey, F. 1973. Grammatical variable in Southwestern American Spanish. Linguistics 108: 5-

---. 1976. Toward a grammatical analysis of Southwest Spanish. In Studies in Southwest

King, J., and J., Marcel. 1991. Individual differences in syntactic processing: The role of

Krashen, Stephen. 1982. Principles and practice in second language acquisition. Oxford:

Larsen-Freeman, D., and Long, M. 1991. An Introduction to Second Language Acquisition

Lee, J. 1987. Comprehending the Spanish subjunctive: an information processing perspective.

Lee, J., and Rodríguez, R. 1997. The effects of lexemic and morphosyntactic modifications on

Lee, J., and VanPatten, B. 1995. Making communicative language teaching happen. New

Leow, R. 1993. To simplify or not to simplify: A look at intake. Studies in Second Language

---. 1995. Modality and intake in second language acquisition. Studies in Second Language

MacWhinney, B. 1982. Basic syntactic processes. Language acquisition: Volume 1: Syntax and

---. 1996. Language specific prediction in foreign language learning. Language Testing, 12, 292-

---. In press. The Competition Model: The input, the context, and the brain. Cognition and

Palmer, F. R. 1986. Mood and Modality. Cambridge, MA: Cambridge University Press.

Pereira, I. 1996. Markedness and instructed SLA: An experiment in teaching the Spanish

Pérez-Leroux, A. 1998. The acquisition of mood selection in Spanish relative clauses. Journal

Pica, T., and Doughty, C. 1985. The role of group work in classroom second language

Pica, T., Kanagy, R., and Falodoun, J. 1993. Choosing and using communication tasks for

Sánchez, R. 1972. Nuestra circumstancia lingüística. El grito 6: 45-74.

Schmidt, R. 1990. The role of consciousness in second language learning. Applied Linguistics

Skiba, R., and Dittmar, N. 1992. Pragmatic, semantic, and syntactic constraints and

Solé, Y. 1977. Continuidad/descontinuidad idiomática en el español tejano. The Bilingual

Stokes, J. 1988. Some factors in the acquisition of the present subjunctive in Spanish.

Stokes, J., and Krashen, S. 1990. Some factors in the acquisition of the present subjunctive in

Swain, M. 1985. Communicative competence: Some roles of comprehensible input and

Terrell, T., Baycroft, B., and Perrone, C. 1987. The Subjunctive in Spanish interlanguage:

Tomlin, R., and Villa, V. 1994. Attention in cognitive science and second language acquisition.

Vainikka, A., and Young-Scholten, M. 1994. Direct access to X'-theory: Evidence from

Vainikka, A., and Young-Scholten, M. 1996. Gradual development of L2 phrase structure.

VanPatten, B. 1990. Attending to content and form in the input: An experiment in

---. 1993. Grammar teaching for the acquisition rich classroom. Foreign Language Annals 26:

---. 1997. The relevance of input processing to second language theory and second language

VanPatten, B., and Cadierno, T. 1993. Explicit instruction and input processing. Studies in

Whitley, M. 1993. Communicative language teaching: An incomplete revolution. Foreign

Woodson, K. 1997. Learner-centered input processing: Bridging the gap between foreign