Barbara A. Lafford (Arizona State University)

Joseph Collentine (Northern Arizona University)

Adam Karp (University of California at Davis)

Comments

After a period of relative neglect vis-a-vis other aspects of second language learning (phonology, grammar and discourse issues), the study of L2 vocabulary acquisition has gained momentum in the last fifteen years. Long and Richards (1997:ix) note that since the mid 1980s there has been "a growing body of empirically based studies of such issues as the nature of the bilingual lexicon, vocabulary acquisition, lexical storage, lexical retrieval, and the use of vocabulary by second language learners." However, most of this research has been carried out on data gathered from learners of languages other than Spanish, e.g., English, German, French as a Second/Foreign Language.

This chapter will present an overview of important issues involved in the study of the acquisition of a second language lexicon by focusing on several key questions:

• Why is the study of second language vocabulary important?
• What items compose the L2 lexicon?
• How is the L2 lexicon organized?
• How is lexical competence determined?
• How is the lexicon acquired (implicit and explicit processes)?

A review of studies of L2 lexical acquisition carried out on data gathered from learners of Spanish as a second/foreign language will be included in the general discussion of these issues. The chapter will conclude with a brief review of research done on strategies for learning L2 vocabulary and an agenda for future L2 vocabulary acquisition research.

The importance of the study of L2 vocabulary is evident from several research findings cited by Gass and Selinker (2001:372): lexical errors constitute most L2 errors and that both learners and native speakers view lexical errors as the most serious and disruptive obstacles to communication. According to Zimmermann (1977:13), Widdowson (1978) claimed: "native speakers can better understand ungrammatical utterances with accurate vocabulary than those with accurate grammar and inaccurate vocabulary." Levelt (1989:181, cited in Gass and Selinker 2001:373) even asserts that the L1 lexicon is the "driving force in sentence production" since it mediates conceptualization and the encoding of grammar and phonology. Gass and Selinker (2001) extend this idea to L2 contexts by stating "In general, there is good reason to believe that the lexicon is an important factor, if not the most important factor, in accounting for he bulk of second language data, in that the lexicon mediates language production." (373)

Many scholars seem to agree that L1 and L2 lexicons are composed of codified lexical items at the word level or higher. According to Schmitt and McCarthy (1997) a lexical item, or a lexeme, is "an item which functions as a single meaning unit, regardless of the number or orthographical words it contains. Fly, pain-induced, and put your nose to the grindstone are all lexical items." (329) That is, the lexicon stores semantic units as both single- and multi-word items (e.g., phrases, idioms, proverbs). Ellis (1997) asserts that among the first items acquired by L2 learners are formulaic utterances ("lexical phrases"), which are later analyzed by segmentation (cf. Nattinger and DeCarrico 1992 and Peters 1983, cited in Schmitt 1997:215), e.g., buenos días, por favor, no sé. Similarly, Terrell’s (1986) Binding-Access framework asserts that unanalyzed phonological chunks are first "bound" to a general meaning and are later re-analyzed into lexical stems and grammatical properties.

Meara (1997) points out that the lack of tradition of using formal models in applied linguistics literature has prevented much fruitful communication between psycholinguistics and applied linguistics. "The L2 research literature contains lots of examples of what might be broadly described as descriptive research on vocabulary acquisition, but very few explanatory, model-based research, which attempts to account for this learning." (Meara 1997:109). Meara, therefore, advocates more model-based research on L2 vocabulary acquisition to facilitate communication between applied linguistics research and other model-based disciplines asking similar research questions about the acquisition of the lexicon. It is therefore proper to ask how formal and cognitive theories of language envision the organization of the lexicon and its interaction with other levels of linguistic representation such as morphology, syntax, and phonology.

The Principles and Parameters perspective envisions the lexicon as a dictionary (Cook, 1988), containing words subcategorized for a variety of syntactic, grammatical and semantic features.

Beber [-noun,+verb], subject=agent, [+transitive]

Bebida [+noun,-verb], [+inanimate]

The lexicon resides in one's peripheral grammar, receiving little assistance from innate (core), universal principles of language. This makes the development of a lexical idiosyncrasies a more difficult task than the development of syntactic knowledge (Cook, 1988). The structure dependence principle (Chomsky, 1988; Cook, 1994) has an important effect on the nature of the lexicon, stipulating that any grammatical operation (e.g., movement, lexical insertion) must consider the syntactic environment at hand. Thus, a learner's lexicon would dictate that bebida (a noun) rather than beber (a verb) could combine with [tengo [...]].

The Minimalist Program (Chomsky, 1990, 1995) posits that all linguistic idiosyncratic reside in the lexicon. In addition to dictionary entries, the lexicon stores derivational, inflectional, and free grammatical morphemes. For instance, Principles and Parameters predicts that learners of an L2 must reset certain parameters. Spanish students might need to learn that the clause is a binding domain; independently, they students would learn the reflexive pronouns. Minimalism posits that one learns individual parameters for individual words. Thus, a student might learn that the referent of se is picked up in the specifier position of the phrase: [ip [d Juan [i' se viste]]; ostensibly, would need to learn the same for me, te, nos, and os. Finally, the lexicon presumably houses functional phrases such as IP and Agr. These abstract lexemes might indicate, for instance, that [estar & -ndo] constitute a constituent or that [ser & ADJ] (without a determiner) is another.

Connectionism aims to model the neural organization and processes of language (cf. MacWhinney, 1997; N. Ellis, 1999). Connectionists assume that the brain stores information in networks of nodes, or relatively discrete knowledge structures. The process of creating (binding) form-meaning relationships occurs when neural networks are strengthened over time as the learner frequently encounters the item in the input. The connectionist perspective predicts that, while different levels of representation are localized (neurologically speaking), there is no strict modularization. From a Saussurian perspective, new form-meaning relationships (linguistic signs) are bound both at the "intrasign" level (e.g., phonological sequence of arbol is bound to the concept of "tree") and at the "intersign" level (e.g., arbor [noun] is related to arbóreo [adjectival form]) to form networks of phonological and semantic associations with other bound forms.

Figure 1. UG and Connectionist Perspectives on Lexical Knowledge.

[place figure 1 here]

Connectionism also views linguistic knowledge to be integrated with academic knowledge (i.e., what we know about the world), be it concrete or abstract (N. Ellis, 1999; Alexander, Schallert, and Hare, 1991). For instance, UG predicts that vocabulary errors result from erroneous lexical entries or performance factors (cf. Levelt, 1989). A connectionist would view an error as an insight into the organization of lexical knowledge and that knowledge's relationship to other knowledge sources. The connectionist perspective recognizes that the words that learners use are quite sensitive to so-called priming effects, or the processing of one node simply because a related node (i.e., one that is semantically or structurally related) has been activated. For example, in an activity relating to food, a learner may be hard pressed to generate in the foreign language a term relating to medicine even if the term is fully acquired, since the learner's present cognitive activity does not activate medical terms.

The connectionist view not only recognizes the existence of codified multiword chunks in the lexicon, but some theoreticians propose a central role for them during production. Crick’s (1979) asserts that the mind has a vast storage capacity but a limited processing capacity (cf. Ellis, 1997:230). Fluency, he asserts, is the use of prefabricated and memorized lexical phrases rather than the employment of syntactic rules. That is, connectionists believe that language production is "the retrieval of larger phrases units from memory" rather than rule governed (Zimmermann 1997:17).

Whereas the UG perspective views lexical development as an accumulation of entities, Connectionism posits that such entities must also be organized into efficient (neurological) networks (Elman et al., 1996). At the moment a learner is discussing (or reading about) foods, he or she should be able to activate culinary terms much more quickly than, say, terms relating to computer technologies.

N. Ellis (1994a/b?) details a connectionist organization of the L2 lexicon. The key organizational principle is the implicit/explicit knowledge continuum. Implicit knowledge is tacit (i.e., difficult to describe) and automatized. It contains knowledge sources that help the learner to encode (i.e., speaking and writing) and decode (i.e., reading and listening) lexical items, termed input/output (I/O) channels. Explicit knowledge is conscious--i.e., available for meta-cognition--and its use requires many attentional resources. This represents the concrete (e.g., a leather, oval ball is for playing football) and abstract (e.g., use usted with people whom we respect) experiences that a language encodes. N. Ellis (1994: 226-7) posits that one's L1 and L2 implicit lexical knowledge stores are localized separately, while one's explicit stores are localized jointly. Presumably, then, a learner's semantic and structural knowledge for any given lexical item is distributed across both explicit and implicit knowledge stores.

How is lexical competence determined?

Meara (1996) proposes that lexical competence is measured by both the size of a learner’s store of lexical items and the organization of those items.

2.3.1 Size

The size of a learner’s vocabulary predicts well a learner's reading and writing abilities (Anderson and Freebody (1981); Perfetti (1985) and Laufer (1998 APPLING :256). Vocabulary size better predicts reading success than syntax and general reading ability Lauffer (1997). Wittrock, Marks and Doctorow (1975) even suggest that a single word can impede comprehension. Nonetheless it is important to note that "a small number of words of English occur very frequently" (Nation and Waring 1997:9). Laufer (1997) conjectures that good L1 readers need a "threshold vocabulary" of about 3000 word "families" (24)--e.g., lengua, lenguas, lenguaje--to interpret an authentic text. Most agree extensive reading, which promotes incidental learning, advances lexical development after the attainment of the threshold (cf. Huckin & Coady 1999: 182). Still, students need strategies for learning less frequent vocabulary after reaching the threshold (Nation 1990).

2.3.2 Organization: What does it mean to know a word?

Knowledge of a word requires an understanding of its spoken and written form, grammatical patterns and collocations, as well as semantic, pragmatic, sociolinguistic, register, connotations (Nation 1990; Nation and Waring 1997). Laufer (1997 blue) discusses factors that make a word hard/easy to learn, such as pronounceability, orthography, morphology, synformy (similarity of form) and semantic features (e.g., abstractness, register restriction, multiple meanings). Henricksen (1999), building on Meara’s work (1996), offers a micro-connectionist view of the L2 lexicon that expands N. Ellis' (1994) macro model. Henricksen distinguishes between three types of knowledge of a word: a. partial/precise knowledge; b. depth of knowledge about the word; c. receptive/productive dimension.

Lexical development involves the incremental "mapping" of various features onto an item. Two processes are involved in "mapping": "semantization/ labeling" and "packaging." The terms "Semantization/Labeling" refer to Saussurian "intrasign" relations, or the mapping of meaning (signatum) onto form (signans). This process is also known as "binding" (Terrell,1986) or, from a connectionist perspective, the strengthening and amplification of related nodes. This process is longitudinally dynamic. A student of Spanish initially might first associate galleta with cracker and later extend it to cookie; or, a learner might over-generalize the extension of pescado to both food and live fish. Several factors may facilitate the semantization/labeling process:

1. Modified input: At times referred to as "teacher / foreigner talk", the essential ingredient here is the simplification of native-like speech or writing (Terrell 1986?; Frey and Grove 199 )

2. Focus on Form instruction: Long (1991) discusses the facilitory effects of teaching content such as geography, art, etc. while drawing students’ attention to particular linguistic features.

3. Noticing: In input processing, forms have the potential to be learned if their attentional resources are directed at that form (Schmidt 1993).

4. Saliency of forms: Noticing is facilitated by frequency of a term in the input, the instructional focus, its communicative importance in a message, phonology, prosody, and part of speech (Brown 19 : 265-66). Unexpected or unusual features also enhance draw the learner's attention to a term (Collentine, 1997; Major, ????; Ryan [1997 S & McC]).

5. Background knowledge and inferring from context: Drawing on one's schemata, prior world knowledge, and the surrounding context (i.e., co-text) promotes form-meaning connections.

6. Depth of Processing: As more nodes are activated during the learning process, the item becomes embedded within a network of nodes instead of residing as a (relatively) isolated node amongst the learner's L2 knowledge store (cf. Stevik 1983).

8. Output Processing techniques: Pushing the learner to produce in the L2 just beyond his/her current level of production promotes binding (o+ 1) (cf. Collentine 1998; Grove 199?; this volume; Salaberry, 199?; see especially Swain 1985, 1983).

"Packaging", or adding features to a lexical item (e.g., pragmatic, sociolinguistic, contextual / dialectic, and metaphoric features), requires extensive reading and listening as well as native-speaker interaction (e.g., study abroad). Packaging is an arduous process. According to Fries (1945 cited in Zimmerman 1997:11) learners often make naïve assumptions about the relationship between the native and target languages e.g., words have exact equivalents in different languages. Upon noticing an L1-L2 disconnect, learners seek other potential meanings (cf. Andersen’s Multifunctionality Principle; cited by Mitchell & Miles 1998:76). Relationships between L1 and L2 lexical forms (cognates) or between similar L2 forms (synforms) can delay packaging. Although cognates are helpful to the L2 learner, false cognates such as embarazada are not. The packaging process may also be affected by the teaching of similar L2 forms (synforms) at the same time, such as "affect/ effect" (cf. Laufer, 1997).

As mentioned above, formulaic chunks (e.g., meter la nariz ' to stick one’s nose into someone’s business ') also behave as discrete lexical items during development. According to Moon (1997) formal errors may result from learner’s failure to recognize that the string of words is non-compositional (e.g., "the smallest" for "at least") or from using a calque of a L1 multiword unit (e.g., Spanish-speakers learning English might use "to pull my hair" [tomarle el pelo a alguien] for "to pull my leg"). Pragmatic errors arise when learners misunderstand or misuse multi-word units in a given discursive context, such as ¡Con permiso¡ 'Excuse me!' after stepping on someone's toe. Stylistic errors may occur "through use of an excessively marked multi-word item—very rare, dated or over-informal- or in an inappropriate genre" (Moon 1997: 60), e.g., using Erase una vez 'Once upon a time' in a historical essay. All of these errors are often exacerbated by the learner's lack of relevant sociocultural background knowledge (e.g., Hay moros en la costa).

Spanish L2 research examining the partial-precise dimension has examined "packaging", studying the acquisition of the two copula verbs (ser/estar) and the prepositions por/para.

The 1970s morpheme order studies explored the sequential development of different morphemes and the stages involved in the acquisition of individual morphemes (cf. Bailey, Madden and Krashen 1974; Dulay & Burt 1974). Spanish L1 and L2 research investigated the Spanish copula relative to other structures (cf. Van Naaerrsen, 1980, 1986). Yet, by comparing these verbs with inflectional morphemes, scholars assumed that the copulas were not lexical phenomena. Abstract semantic features (primarily relational features, with estar being more marked than ser), nevertheless, distinguish the two (Lafford 1986).

Van Naerssen (1980) used oral data to study copula acquisition, revealing that students had good control (73.7% accuracy) after a year of study using the Natural Approach. However, the accuracy rates for ser and estar were averaged into one category "copula" (due to a coding error). Smith (1980), whose subjects were at the ??? level, also found good "copulas" control in compositions; however, these scores may be inflated since the participants could monitor production.

In the 1980s scholars shifted their attention from orders of acquisition (recognizing the limitations of the 'accumulated entities assumption'; cf. Dulay, Burt and Krashen 1982) to stages of acquisition of particular structures (e.g., negation, WH questions). Van Patten (1985, 1987) investigated the stages of ser and estar. His 1985 study involved six beginning level, university-level adults learning Spanish as a foreign language in a communicatively-based classroom. Tasks involved conversations and picture description/story telling. The 1987 study added a grammaticality judgment task with a different group of learners (with one to two and a half years of study) and classroom observation data. VanPatten posits five--largely sequential--stages.

I Absence of copula in learner speech

*Juan alto . (John is tall.)

II. Selection of ser to perform most copula functions

Juan es alto.

*Juan es enfermo. (John is sick.)

*Juan es estudiando. (John is studying.)

III. Appearance of estar with progressive.

Juan está estudiando.

IV. Appearance of estar with locatives.

Juan está en la clase. (John is in the classroom.)

V. Appearance of estar with adjectives of condition

VanPatten (1987) proposes that the interaction of four developmental processes accounts for these stages: (1) Communicative value -- both copulas are "meaningless functors" in terms propositional content. (2) Simplification -- in stage II, the learners used only one of the copulas. Additionally, the non-contrastive uses appear (e.g., estar + progressive) before the contrastive uses (e.g., ser/estar + locatives, estar + adjectives); (3) Transfer -- English-speaking learners tend to linger in stage II since English has only one copula (cf. Andersen’s "Transfer to Somewhere Principle"); (4) frequency in input -- learners hear and read less estar forms.

De Keyser (1990) showed that the classroom development differs from study-abroad development and ser is used more accurately than estar during this process.

Ryan and Lafford (1992) examined the stages of ser and estar in a study-abroad context. Sixteen students in an intensive beginning-level, study abroad program participated in a series of oral interviews. Ryan and Lafford defined "acquisition" as 90% accuracy on the use of a copula in a given function on at least one of the interviews. Like VanPatten, they found ser use consistently more correct than estar; ser accuracy hovered around 90% throughout the data-collection period while estar generally never reached that level.

Ryan and Lafford proposed the following eight stages:

1. Absence of the copula

2. Overuse of ser in estar Locative and Conditional contexts.

3. *Absence of the copula in estar Conditional contexts.

4. Estar in progressive constructions.

5. *Replacement of zero copula errors by errors with estar where ser is required.

6. *Estar replaces zero copula in Conditional contexts

7. Estar in Conditional contexts

8. Estar replaces ser in Locative contexts.

Ryan and Lafford (1992) propose stages (#s 3, 4, and 6) in addition to VanPatten's and partially contradict his findings: estar is acquired in Conditional before Locative contexts; they observed a protracted use of zero copula in Conditional contexts calling for estar. The input learners receive abroad may influence these stages. Ryan and Lafford conjecture that store signs such as Abierto and Cerrado may confuse learners about which copula is required with conditionals. [I couldn't think of a concise way to describe the "filling-the-gap" hypothesis.]

Guntermann (1992a) investigated ser and estar stages, examining conversational data (Oral Proficiency Interviews) from nine Peace Corps volunteers who studied Spanish in both the classroom and abroad. She analyzed two conversations: one completed at the completion of the experience abroad and another a year later. Guntermann's findings were partially consistent with VanPatten (1985, 1987) and Ryan and Lafford (1992), adding a stage: the late acquisition of the ser passive construction. Guntermann also provides a functional perspective, discussing the discursive roles of certain formulaic expressions (e.g., es muy mal as a topic closer).

Finnemann (1990) investigated the interaction of markedness relations between ser and estar and learning styles (form- or meaning-oriented) and copula development. He charted the copula development of three university-level, foreign-language learners; at the inception of the study, neither had any Spanish knowledge. Based on oral data, Finneman argued that learning style interacted with how these learners employed the two copulas.

Guntermann (1992b) used the same methodology as Guntermann (1992a) to investigate the acquisition of the prepositions por and para. Lunn (1978) informed the data codification (with some modifications) and native speakers largely corroborated the analysis. Excluding formulaic utterances (e.g., por ejemplo), subjects were significantly more accurate with para (73%) than with por (32%); para was targeted more frequently than por. Guntermann (1992b) offers several explanations: 1) the Naive Lexical Hypothesis, or every L1 word has an L2 equivalent, causes learners to equate L1 "for" with L2 por; 2) para is less marked than por (cf. Beale 1978); 3) and, similar to VanPatten's (1985) communicative value hypothesis, para seems to be more essential for communication from the beginning. Furthermore, similar to Finnemann (1990), individual learners tend to favor one preposition, with por being favored by less proficient learners.

Lafford and Ryan (1995) investigated the correct suppliance of por and para and overgeneralizations of learners (N = 9) abroad. The data were from ACTFL-style oral interviews-- three times throughout one semester --and two native speakers corroborated the analysis. Lafford and Ryan propose the following relative order of acquisition of various "canonical" functions of por and para :

Por-functions acquired early

1. expressions: por favor

2. duration: por… un ano y medio

3. action within a timespan: Hago ejercicios por la manana.

Por- functions acquired next

4. exchange: y por dos habitaciones, ¿cuánto cuesta?

5. motive : cuatro Oscars por mejor película

6. movement through space: recto por la Calle Ronda

Por- functions acquired late

7. movement within a space: doy un paseo por la ciudad

------------------------------------------------------------------------------------------------

Para- functions acquired early

1. purpose: para levantar pesas

2. standard: es más fácil para nosotros

3. beneficiary: es para caballos

Para- functions acquired next

4. intended use: caja para gasoline

5. landmark: voy para la casa

6. deadline : tiene habitaciones para este fin de semana

Para- functions acquired late

7. para que + subjunctive: para que los hijos pueden…

Lafford and Ryan (1995) noted that, consistent with Andersen’s "One-to-One Principle," para consistently conveys the notion of "goal". Interestingly, this learner hypothesis extends to pragmatic uses, as para was even used as a deictic marker of topicalization: ¿Como son diferentes o semejantes tu casa en Arizona y tu piso aqui? "How are your house in Arizona and your apartment here different or similar?" one informant responded Para mi casa en Arizona, es mismo de este casa. ([As] for my house in Arizona, it is the same as this house).

Liontas (1999) is the only study examining multiword-utterances, examining the comprehension and interpretation of vivid phrasal idioms, whose meaning is not transparent based on its elements (e.g., le falta un tornillo= "he has a screw loose"; cuando las ranas críen pelo _"when frogs grow hair" = "when pigs fly"). Eighty English-native foreign-language learners of French, German and Spanish participated in the study, all enrolled in third-level university classes participated in this study. Experimental and qualitative data indicated that a learner’s lack of background and cultural knowledge interacts with comprehension: "the degree of difficulty associated with the comprehension and interpretation of the TL [target language] idiom corresponds to the degree of distance associated with the semantic/imagery distance between TL and NL idioms." (119)

b. Depth of Knowledge dimension The least studied aspect of the learner's lexicon is the depth-of-knowledge dimension (Henricksen 1999), or the structure of the student's lexical knowledge. Learners need optimal neurological networks to access them efficiently, achieved through the creation of "intensional links" and "sense relations" between lexemes (cf. Aitchison 1994; Miller and Fellbaum, 1991). For example, ardiente might be linked to caliente via a synonymous connection, despierto and dormido via antonymy, and fresco and frío as a gradation. Markedness relations also influence the network-construction process. Early on, errors and word-association tasks indicate that the intrasign connections between synforms, or phonologically similar signs, have the most robust influence on performance (cf. Meara, 1978; Gass and Selinker 2001). Later on, grammatical relations have an important impact on performance. Collentine and Lafford (1987) report that third-year FL learners regularly commit part-of-speech errors, e.g., the use of Voy a la universitaria 'I am going to the (pertaining to the) university (adjectival meaning).' instead of universidad 'university (noun form)'. Learners apparently are quick to incorporate a new term into their networks which might fill some perceived gap (Brown et al. 1999). In the absence of a known candidate, students often employ an L1 word or create neologisms, e.g., the use of tení for tuve (Lafford and Collentine, 1987).

Lafford & Collentine (1987) oral-production data from third-year, university-level students (N=???) enrolled in a proficiency-based program. Using a Jakobsonian semantic framework, they outline intersign relationships that may exist in a learner’s interlanguage, assuming that signs are interconnected through four basic relationships involving "similarity" (involving the notion of resemblance, e.g., simile, metaphor) and "contiguity" (entailing temporal or spatial adjacency, cause/effect, part/whole).

1. formal partners: those signs related only via similarity relations among signantia [Laufer’s ‘synforms’] with no obvious semantic similarity or contiguity connections (e.g., sentar/sentir).

2. grammatical paradigms: those signs which constitute a grammatical category and are related primarily through similarity relations of markedness among grammatical features in the signata of different signs, e.g., the morpheme indicating number in the determiners los and las.

3. lexical/derivational constellations: those signs related by similarity relations among the signantia and among lexical features in the signata , e.g., universitaria vs. universidad.

4. semantic schemata: groups (semantic fields) of grammatical paradigms and lexical/derivational constellations whose signata are related through relations of similarity and contiguity, e.g., all terms related to "writing" such as carta, pluma, papel, escribir, escritura, escribo.

Ife et al. (2000), asking whether greater proficiency correlates with greater vocabulary learning abroad, studied intersign relationships amongst L2 Spanish learners. University-level learners (N = 36) participated in a word-association task and a translation task before and after their study-abroad experience (ranging 4-8 months). No significant results emerged, although trends appeared indicating gains in both size and lexical organization for both intermediate and advanced learners. Gains in lexical organization were especially noted in the Advanced learners who stayed abroad for a longer period of time (two semesters = 8 months).

Henrickson’s (1999) third dimension is consonant with the assumptions of several scholars (including Teichroew [1982, cited by G&S 2001:375; Palmberg [1987]; Gass [1988]; Laufer [1990] and Melka [1997]) that lexical knowledge is best conceptualized as a continuum between the ability to recognize the meaning of a lexical item and the ability to use it productively, effectively rejecting the popular conception that one possesses a receptive and another productive set of vocabulary items. To increase the likelihood that a term will be available for production, learners must practice retrieving L2 items, which strengthens and automatizes so-called I/O channels (i.e., knowledge for decoding and encoding structural properties; Ellis 1994). Bialstok and Sharwood Smith (1985) contend that learners need "control" over--as opposed to mere "knowledge" of--a lexeme's use for it to appear in output. Laufer and Paribakht (1998) investigate what they term passive, controlled active and free active vocabulary knowledge. Each develops at a different rate: passive is fastest, free active is slowest and controlled active occupies an intermediate position between these two (cited in Gass & Selinker 2001:375). To these authors' knowledge, no research has been conducted to date on the receptive / productive dimension in Spanish L2 contexts.

Other developmental factors that have not been fully discussed in the Spanish L2 investigations to date surely interact with all three of Henricksen's dimensions. First, research on "Cross Linguistic Influence", which amounts to a resurgence in the interest in what has been termed 'transfer, indicates that the L1 is an important factor in the development of L2 vocabulary (cf. Kellerman and Sharwood Smith 1986 cited in G&S 2001?:. 89). Bowen & Stockwell’s (1965) behavioristically- based Hierarchy of Difficulty predicts that substantial problems will occur when the L2 contains two semantic categories for which the L1 has just one form. Thus, native speakers of English often have trouble distinguishing the copulas or the prepositions por/para since both pairs can mean 'for'. Additionally, the L1 conspires to keep learners in a certain stage of acquisition of the L2 morpheme if performance is similar L1 performance. English-speaking students may continue to rely on ser for most copula functions for a long period of time since es (ser) is so similar to English is. (See G&S p. 99. Kellerman 1979 for issues of coreness and that similarity of L1 and L2 determine the amount of transfer. G&S p. 102-103. See also Swan blue book] on the influence of L1 on L2 acquisition.). Second, Major’s (2001) Similarity-Difference Rate Hypothesis states that it is easier to acquire L2 forms that are truly different from each other. Therefore, ser/estar and por/para are harder to acquire because they are so similar in form and meaning, not because they are so different.

Little is known about the acquisition of vocabulary (and much less Spanish vocabulary, to which the previous section testifies), and clear-cut principles with which to approach the teaching of vocabulary are scant. In her review of historical trends in L2 vocabulary instruction Zimmermann (1997) states: "although the lexicon is arguably central to language acquisition and use, vocabulary instruction has not been a priority in second language acquisition research or methodology." (17) The most substantive help that instruction has received has been the efforts of scholars to collate word-frequency lists. Yet, scholars now recognize that instruction has generally assumed that vocabulary acquisition proceeds "implicitly" (cf. N. Ellis 1994) or incidentally (Gass 1999). Indeed, Krashen and Terrell (1983) affirm: "Comprehensible input is the essential environmental ingredient and a richly specified language acquisition device (LAD) assimilates vocabulary from the evidence provided in natural language." (quoted in Ellis 1994b:212 ) Yet, N. Ellis (1994) affirms that vocabulary learning is greatly enhanced by explicit instructional intervention, or techniques that promote the deep processing of terms (i.e., using and process a term in a variety of contexts to better incorporate it into the learner's lexical network). In this regard, research into learning strategies is beginning to provide a plethora of techniques and assumptions for pedagogues and materials designers to consider.

Word frequency lists contain "core vocabulary which L2 and L2 teachers are recommended to use to decide which words and meanings should be taught first." (N. Ellis 1994b, p. 212). West’s (1953) list of high frequency words in English is still used in spite of more current lists created through computer technology (Meara 1980 cited in Zimmerman 1997:9). With the arrival of more communicative teaching methods, the use of frequency lists has recently been criticized by some scholars (See Zimmermann 1977:14 for a review of this research), although some still find them useful for curriculum design (Nation and Waring 1997:17-18).

Buchanan (1927) and Keniston (1941) created lists for curriculum designers and teachers of Spanish. However, Bull (1949, 19 ) questioned the lists' reliability, wondering whether the same lists would arise from samples of a variety of materials. He argued that there was very little predictive power in Keniston’s list outside of pedagogical texts, since the words were rare in a set of authentic literary texts that Bull and his assistants sampled. Despite the creative contributions to the field of Spanish Applied Linguistics that Bull would make later on (1965), this 19 study shows that he had not yet broken away from the 19thC tradition on relying on literary texts for descriptions of the Spanish language. Due to fact that scholars have found that written discourse evidences a greater variety of lexical items than oral discourse (Chafe and Danielwicz 1987; Biber 1988, 1995 cited in Brown et al. 1999:265), it is possible that frequency lists based on transcriptions of oral data would yield very different results.

Ellis (1994b) contends that mapping form to meaning is the result of cognitive mediation which depends upon explicit learning processes, e.g., mnemonics, semantic or imagery mediation between the FL word (or a keyword approximation) and the L1 translation. Fraser (1999:226) found positive results for the teaching of so-called Lexical Processing Strategies (LPS). Coady (1997) notes that the role of learner strategies vocabulary instruction approaches strategies identified four major positions toward the use of strategies for vocabulary instruction: 1) No formal strategy use: vocabulary acquisition occurs in context as the learner engages in extensive reading (Krashen [1985-89]), 2) Strategy instruction: formal instruction in the use of vocabulary learning strategies is essential to learning items in context (Oxford and Scarcella [1994]) 3) Development plus explicit instruction: a combination of formal instruction and vocabulary learning strategies is needed (Paribakht and Wesche [1997] and Zimmermann [1994]) 4) Classroom activities: this approach advocates the teaching of vocabulary along traditional lines, often using handbooks which emphasize practical activities (Allen [1983] suggests different types of activities for beginning, intermediate and advanced learners).

3.2.1 Guessing from context: A common L2 vocabulary learning strategy involves inferring a lexeme's meaning from it oral or written context (Haastrup 1991 cited in Paribahkt &Wesche 1999:199). Inferring involves the consideration of graphomorphemic (orthographic), morphological (e.g., roots, stems and affixes), syntactic and semantic qualities; it also entails the use of one's background knowledge (Lee and Wolf 1997). The research suggests that beginning L2 readers rely on graphomorphemic correspondences, a bottom-up strategy (Coady 1979; Lee and Wolf 1997). Higher level learners may be more effective guessors because they possess a larger lexicon, which in turn allows them to employ both bottom-up and top-down strategies. Interestingly, however, Mondria and Wit-de Boer (1991:262 cited in Clipperton 19 :743) hypothesize that, if context makes meaning too clear, lexical retention diminishes.

Lee & Wolf (1997) examined lexical inferencing strategies used by Spanish native speakers and university-level native English-speaking learners (N = 28) of Spanish at beginning, intermediate and advanced levels. They employed a retrospective think-aloud protocol, asking participants to describe how they guessed the meanings of certain words. The results indicate that inferences increase in acceptability with greater proficiency and less proficient learners favor "bottom-up" strategies while advanced learners (and natives) favor "top-down" strategies.

Pulido (2000) looked at the impact of topic familiarity, reading proficiency, sight vocabulary (i.e., one's overall knowledge of Spanish vocabulary in the task), level of instruction (beginning, intermediate and advanced university-level study) on incidental vocabulary gain through reading. The 99 participants read four brief narratives with embedded nonsense words. Based on two comprehension tasks, she found that L2 reading proficiency was the strongest predictor of gain. Topic familiarity may also be a predictor, accounting for gains on one of the tasks.

Foreign language reading materials have long glossed texts largely with English translations and paraphrases / definitions in the target language. Scant research has examined the actual effects of glosses on comprehension or retention.

Hulstijn (1992) found that the mental effort associated with a gloss (e.g., selecting one translation from a series of four) best predicted comprehension. Additionally, Hulstijn found that learners are more likely to incorrectly infer the meaning of an unknown L2 word if no cue as to the word's meaning is provided than when a cue has been given. Watanabe (1997) concludes that more natural glosses such as appositives are not more effective than glosses such as single-word or multiple-choice translations.

Jacobs, Dufon and Hong (1994) investigated the effects of glossing on recall and retention with a written recall protocol and a translation task. University-level, native speakers of English (N = 85) at various levels of proficiency studying Spanish as a foreign language participated, reading a text either with no glosses, English glosses, or Spanish glosses. The recall data suggested that glossing affected recall only at higher levels of proficiency. Participants reading the unglossed text outperformed other learners on the translation task, although this difference was not sustained over time.

3.2.3 The Use of Dictionaries:

Researchers have wondered whether dictionary use during reading tasks affects on vocabulary learning. The studies examining the effects of dictionary use on vocabulary learning have involved bilingual rather than monolingual dictionaries. Luppescu and Day (1993) report that dictionary use contributes to L2 vocabulary learning although it does slow reading and may cause confusion when an entry has more than one definition.

Two investigators have compared dictionary use to alternative means for defining potentially indeterminable lexemes. Hulstijn, Hollander and Greidanus (1996) show that dictionary use leads to greater comprehension than marginal glosses and reoccurrence of unknown words. Additionally, Fraser’s (1999) study states that inference tasks followed by dictionary use affects vocabulary learning.

Knight (1994) reports on an experiment that looked at the effects of intermediate level (second year) Spanish learners' (N = ???) verbal ability level (based on ACT scores) and bilingual dictionary (specifically, an electronic, on-line dictionary) use on incidental vocabulary learning and reading comprehension. The researcher controlled for dictionary use and the subjects overall verbal ability. All told, the students learned more vocabulary when using a dictionary, and high verbal ability students learned more words than low verbal ability students. Additionally, access to the dictionary allowed low level students to gain receptive knowledge of almost as many words as the high level students.

(Get other studies from Diana Chin’s thesis on this.) Coady 1997 p. 286.

The increasing availability of computers and electronic corpora in recent years has facilitated the creation of new types of tools, such as on-line dictionaries and even inverse dictionaries (Teschner 19 new?). Recognizing that students learn using different cognitive styles, Chun and Plass (1996a, 1996b) investigated the effects of different multimedia modalities on incidental vocabulary learning by native English-speaking learners of German. Their studies partially support the "dual-coding effect": a word annotated or coded with combined textual + visual or audio + visual modes of information will be learned better than a word coded with only the textual or audio mode (Paivio, 1986). Another interesting finding by Chun and Plass (1996) involves the need for repetition and manipulation of the new L2 items in order for them to be acquired.

Kang (1995)?

Karp’s research (2001) Joe: I’m still waiting for this.

3.5 Semantic mapping. Semantic mapping involves the building of diagrammatic maps which illustrate how certain word clusters are associated with a key word, idea or concept. This technique was originally developed by Johnson and Pearson (1978) to teach vocabulary to children learning to read in their native language. (Omaggio 2001:376.).

Morrin and Goebel (forthcoming) study the effects of semantic mapping--clusterings based on semantic and syntactic similarities (e.g., apricot, peach, plum, nectarine, pear, apple)--on the acquisition of Spanish L2 vocabulary by English-speaking college students (N = ???). A control group employed a set of vocabulary items in communicative tasks (e.g., small group and pair assignments). An experimental group participated in the same tasks, as well as formal vocabulary instruction and were also taught semantic mapping strategies. The results show that although both groups knew approximately the same amount of vocabulary (providing L1 definitions for a list of L2 words), the semantic mapping group remembered having heard more words on the list provided to them, implying that the effects were on phonological links. The study also indicates that the semantic mapping group significantly outperformed the vocabulary activities group in meaning recall and in their ability to organize L2 vocabulary according to thematic relations to other words.

Investigations into vocabulary learning is just beginning to become an important part of the SLA research program. As in any new area of study, wide ranging discrepancies in research design make it difficult to make sweeping generalizations about the potential implications of this research for the classroom. It is important to note that many of the researchers cited above do, in fact, acknowledge these shortcomings.

Three methodological dearths limit the extent to which findings are generalizable to the population of Spanish students as a whole. First, a recurrent shortcoming of the studies reported here stems from the small pool of subjects that data collection tasks sampled (e.g., De Keyser 1990; Finneman 1990; Guntermann 1992b; Ife et al. 2000; Lee & Wolf 1997; Ryan and Lafford 1992, 1995; VanPatten 1985, 1987). A notable exception is De Keyser (1990), employing three data collection tasks. Second, the proposals that learners advance through certain stages (Guntermann 1992b; Ryan and Lafford 1992, 1995; VanPatten 1985, 1987) certainly requires corroboration via longitudinally, either with a highly qualitative study of a small sample or a cross-sectional study realized over an extended period of time. Another approach to this dilemma could entail a cross-sectional multivariate study (examining a number of variables with a relatively large sample) and submitting the data to a path analysis (a statistical exploratory technique). Third, conclusions drawn from a single data-collection technique naturally demand tenuous interpretations about the status of the learner's development; any task places particular processing demands on the learner which in turn influences performance (Guntermann 1992b; Pulido 2000; Ryan and Lafford 1992, 1995; VanPatten 1985, 1987). For instance, distinct orders may have surged in the ser / estar and por / para stage studies with, for instance, composition data. Recall protocol studies are hard pressed to reveal subconscious processes, masking any implicit learning that may have occurred.

Future investigations should pay closer attention to the construct validity of their methodologies. Such validity can be corroborated with more explicit descriptions of one's data analysis For instance, VanPatten (1985, 1987) does not clearly operationalize "obligatory" contexts for the use of ser / estar. [Bobbie, does your research do so?] Notable exceptions are Guntermann (1992b), delineating clear a theoretical thesis for her data analysis, and Lafford and Ryan (1995), who recruited native speakers for corroboration. Another potential limitation in this research relating to construct validity is Pulido's operationalization of topic familiarity, which may vary greatly between subjects even when the topic is mundane.

Other shortcomings in the above-mention research relate to nature and the number of instruments employed. First a major limitation noted of Jacobs, Dufon and Hong (1994) concerns the fact that the Spanish glosses were not pilot tested to see whether the students could understand them. Second, Ife et al. (2000) employed no control group, such that it is unclear the extent to which the subjects simply improved in terms of their ability to complete the data-collection task. Finally, both Knight (1994) and Morrin and Goebel (forthcoming) designed studies in which the treatment groups simply received more time on task, which may significantly account for their findings.

Finally, a common theme in all SLA research is the extent to which a study measures comprehension or intake; that is, are observed gains long term. These are clearly separate issues and their consideration has an important impact on pedagogical recommendations.

Approaches to the analysis of L2 vocabulary data have been varied. While most L2 lexical studies to date have been based on written data, some studies have been carried out using oral data. Some scholars have preferred carrying out quantitative studies (e.g., Nation and Waring 1997) to establish a core L2 vocabulary based on calculating the most frequent words used in the L2, while others have taken a more qualitative approach to see how words are learned. (e.g., Parry [1997], Altman [1997] and Grabe and Stoller [1997]). Peribahkt and Wesche (1999) used talk aloud protocols and retrospective protocols to get at learners’ strategies to acquire L2 vocabulary.

Based upon this review of the literature, it is clear that there are several issues regarding L2 vocabulary acquisition that still need to be addressed. Status of the lexicon under the Minimalist Program is that it is a store of all idiosyncratic L2 information, lexical and grammatical. Additionally, researchers approach the acquisition of Spanish phenomena with varying assumptions about the grammatical and / or lexical status of phenomena such as ser / estar and por / para. Clearly, then, the most important issues for researchers to address now are the following:

• What are the limitations of the theoretical models delineated here?

• Do certain heretofore assumed grammatical phenomena behave more like lexical phenomena during development (i.e., their incorporation into the learner's interlanguage)?

• To what extent do the theories of lexical development and organization detailed above account for overall L2 development better than theories of grammatical development?

At some point, a extensive proposal for the "methodology" of vocabulary instruction will be welcomed by instructors and materials designers. Such a methodology (and its inevitable shortcomings) should provide as much predictive power as those that aim at the development of grammatical knowledge (e.g., processing instruction).

Researchers also would do well to examine the acquisition of other apparent lexical obstacles for learners of Spanish, such as saber / conocer? Such studies would provide further insights into the factors that interact with the processes of mapping, packaging and depth of knowledge. Clearly, a better understanding of the receptive / productive dimension as it relates to the acquisition of Spanish is necessary as well.

What are the relative effects of different types of input modifications on vocabulary learning through inferred meaning? That is, how effective are particular types of glosses, cues, or tasks for allowing learners to infer the meaning of unknown words without running the risk of making incorrect inferences? Furthermore, does the interaction associated with computer-mediated tasks lead to an increase in depth of word knowledge and retention of meaning without distracting learners from comprehending what they read? Finally, a benchmark is needed for how much vocabulary knowledge changes over time in advanced Spanish students with respect to various types of glosses, cues, and tasks. (I didn’t write this last part, I think Karp did.)

Although there has been an advantage in exploring various ways to describe and measure L2 vocabulary acquisition, the lack of a coherent research agenda has prevented the field from developing at a faster rate. As stated earlier, Meara (1997) proposes that future research in vocabulary acquisition should involve the use of common models and more communication with researchers in other disciplines. He also advocates for researchers trying to ask the same questions, rather than having each one follow a different line of research, with no replication of studies or accumulation of knowledge in given areas. As the twenty-first century dawns, L2 research scholars need to answer this call for more rigor and systematically in the study of the acquisition of second language lexicons.

Works Cited

Alexander, Patricia, Diane Schallert, and Victoria Hare. 1991. Coming to terms: How

researchers in learning and literacy talk about knowledge. Review of Educational

Research, 61, 315-343.

Buchanan, Milton A. 1927. A Graded Spanish Word Book. Toronto.

Bull, William. 195 . " Spanish Word Counts: Theory and Practice." (use Hispania CDROM to find this.

Chomsky, Noam. 1988. Language and problems of knowledge: The Managua lectures.

Cambridge, MA: The MIT Press.

Collentine, Joseph G. 1997. The Effect of Irregular Stems on the Detection of Verbs in

the Subjunctive. Spaniah Applied Linguistics 1.1:3-23.

Cook, Vivian. 1988. Chomsky's Universal Grammar: An introduction. Oxford:

Blackwell.

Ellis Nick. 1994a. Universal Grammar and the learning and teaching of second

languages. In Perspectives on pedagogical gramma, ed. Terence Odlin, 25-48.

Cambridge: Cambridge University Press.

-----. 1994b. Vocabulary acquisition: The implicit ins and outs of explicit

cognitive mediation. In Implicit and explicit learning of languages, ed. Nick Ellis,

211-282. New York: Academic Press.

Ellis, Nick. 1997. Vocabulary acquisition: word structure, collocation, word-class, and

meaning. In Vocabulary: description, acquisition and pedagogy, ed. Norbert Schmitt

and Michael McCarthy, 122-39. Cambridge: Cambridge University Press.

-----. 1999. Cognitive approaches to SLA. Annual Review of Applied Linguistics, 19,

22-42.

Elman, Jeffrey. 1990. Finding structure in time. Cognitive Science, 14, 179-211.

Elman, Jeffrey L., Bates, Elizabeth A., Johnson, Mark H., Karmiloff Smith, Annette,

Domenico, Parisi, and Kim Plunkett. 1996. Rethinking innateness: A connectionist

perspective on development. Cambridge, MA: The MIT Press.

Fodor, Jerry. 1983. The modularity of mind: An essay on faculty psychology.

Cambridge, MA: The MIT Press.

Gass, Susan and Larry Selinker. 2001. Second language acquisition: an introductory

course. 2d ed.. Mahwah, NJ: Lawrence Erlbaum Associates.

Ife, Anne, Vives Boix, Gemma and Paul Meara. 2000. "The impact of study abroad on

the vocabulary development of different proficiency groups." Spanish Applied

Linguistics 4.1:55-84.

Keniston, Hayward. 1941. A Standard List of Spanish Words and Idioms. New York.

Lafford, Barbara. 1986 "Ser y estar: una nueva dimension." Actas…….SER/ESTAR

paper on markedness. (Mexico City)

Lafford, Barbara A. and Joseph G. Collentine. 1987. "Lexical and Grammatical Access

Errors in the Speech of Intermediate/Advanced Level Students of Spanish."

Lenguas Modernas 14:87-112.

Lee, James, and Darlene F. Wolf. 1997. "A Quantitative and Qualitative Analysis of the Word-Meaning Inferencing Strategies of L1 and L2 Readers." Spanish Applied Linguistics 1.1: 24-64.

Leow, Ronald. 1995. Modality and Intake in Second Language Acquisition. Studies in

Second Language Acquisition. 17:79-90.

Levelt, Willem. 1989. Speaking: From intention to articulation. Cambridge, MA: The

MIT Press.

Liontas, John I. 1999. Developing a Pragmatic Methodology of Idiomaticity: TheComprehension and Interpretation of SL Vivid Phrasal Idioms During Reading. Ph.D. dissertation. University of Arizona.

Long, Michael and Richards, Jack C. 1997. Series editors’preface. In Second language

vocabulary acquisition, ed. James Coady and Thomas Huckin, ix-x. Cambridge:

Cambridge University Press.

MacWhinney, Brian. 1997. Second language acquisition and the Competition Model. In

Tutorial in bilingualism: Psycholinguistic perspectives ed. Annette de Groot and Judith

Kroll, 113-142. Mahwah, NJ: Lawrence Erlbaum Associates.

Meara, Paul. 1997. Towards a new approach to modeling vocabulary acquisition. In

Vocabulary: description, acquisition and pedagogy, ed. Norbert Schmitt and Michael

McCarthy, 109-21. Cambridge: Cambridge University Press.

Moon, Rosamund. 1997. Vocabulary connections: multi-word items in English. In

Vocabulary: description, acquisition and pedagogy, ed. Norbert Schmitt and Michael

McCarthy, 40-63. Cambridge: Cambridge University Press.

Nattinger, J.R. and J.S. DiCarrico. 1992. Lexical phrases and language teaching. Oxford:

Oxford University Press.

Schmitt, Norbert. 1997. Vocabulary learning strategies. In Vocabulary: description,

acquisition and pedagogy, ed. Norbert Schmitt and Michael McCarthy, 199-227.

Cambridge: Cambridge University Press.

Schmitt, Norbert and Michael McCarthy. 1997. Glossary. In Vocabulary: description,

acquisition and pedagogy, ed. Norbert Schmitt and Michael McCarthy, 327-31.

Cambridge: Cambridge University Press.

Sinclair, J. 1985. Selected issues. In English in the world, ed. R. Quirk and H.G.

Widdowson, 248-54. Cambridge: Cambridge University Press.

Terrell, Tracy. 1986. Acquisition in the Natural Approach: the Binding/Access

framework. Modern Language Journal. 70.3:213-27.

Van Naerssen, Margaret. 1980. "How Similar are Spanish as a First Language and Spanish as a Foreign Language." Research in Second Language Acquisition. Eds. Robin C. Scarcella and Stephen D. Krashen. Rowley, MA: Newbury House. 146-54.

---. 1986. "Hipótesis sobre la adquisición de una segunda lengua, consideraciones inter-lenguaje: comprobación en el español." Ed. Jürgen Meisel. Adquisición de lenguaje: Adquisicao de linguagem. Frankfurt: Vervuert. 139-55.

Zimmerman, Cheryl Boyd. 1997. Historical trends in second language vocabulary

instruction. In Second language vocabulary acquisition, ed. James Coady and Thomas

Huckin, 5-19. Cambridge: Cambridge University Press.

--------------------------------------------------------------------------------------------------------

Aitchison, Jean. 1994. Words in the mind: An introduction to the mental lexicon. 2nd

edition. Oxford: Blackwell.

Henriksen, Birgit. 1999. Three dimensions of vocabulary development. Studies in

Second Language Acquisition, 21, 303-318.

Meara, Paul. 1996. The dimensions of lexical competence. In. Gillian Brown, Kirsten Malmkjaer, & John Williams (Eds.), Performance and competence in second language acquisition (pp. 35-53). Tampere, Finland: Publications de l'Association Finlandaise de Linguistique Appliquée.

Miller, George A. and Fellbaum, Christiane. 1991. Semantic networks in English. Cognition, 41, 197-229.

Wesche, Marjorie, & Paribakht, T. Sima. 1999. Incidental L2 vocabulary acquisition: Theory, current research, and instructional implications (Introduction). Studies in Second Language Acquisition, 21, 175-180.