The art of being wise is knowing what to overlook.
—William James, American psychologist and philosopher The Principles of Psychology
(1890:Vol. 2 Ch. XXII: Reasoning. Different Orders of Human Genius)
What is a ‘pedagogical grammar’ (PG)? Like grammar, this term is ambiguous. Corder (1975a), for instance, understands the concept as “the statements about, and exemplifications of, the language which are for the use of teachers rather than learners, the object of which is then to guide the teacher in the way he is to present the language material to his pupils.” In other words, for Corder PG is for pedagogues. PG denotes language descriptions geared to the demands of teaching; the pedagogical linguist will shape his grammar according to the demands of the given teaching situation/s (Sharwood Smith 1974/81:17). In order to avoid potential misunderstandings concerning this term it is expedient to follow a basic division of grammar into four kinds, with the nomenclature suggested by Professor Romuald Gozdawa-Gołębiowski (p.c.).
An author intending to write an ideal grammar book would typically wish to begin with a mechanism (system) representing native speakers’ competence of their language. This kind of grammar is called ‘internalised grammar’ (IG; also labelled ‘intentional grammar’ or ‘mental grammar’ in some sources). However, the notion of competence, which was defined by Chomsky (1965) as the inborn, tacit and subconscious ‘all-time’ steady-state knowledge of language rules not subject to introspectional investigation, raises some questions. Hence, it is more appropriate to talk about performance (the ‘real-time’ use of that knowledge in the comprehension and production of utterances; “the actual use of language in concrete situations;” op. cit.:4), which is believed to be the reflection of the underlying competence in actual communication, from which the mental grammar can be inferred.
The output of IG, its linguistic model, a scientific description of the formal language system in the shape of a set of rules allowing users to construct well- but not ill-formed sentences, additionally containing theoretical assumptions concerning the nature of language and these days obligatorily verified by insights from corpus research, is referred to as ‘externalised grammar’ (EG). It also called descriptive or linguistic grammar (i.e. “a description of language forms, and in some cases functions, cast in a coherent, constrained, and self-contained metalanguage;” Tomlin 1994:143). EG is based on metalingual knowledge: highly analysed knowledge reflected in the speaker’s ability to denominate categories and verbalise rules using grammarians’ jargon (when described by grammarians, it has the form of a reference grammar).
EG is the source from which teachers-practitioners select, draw and subsequently adjust rules to meet the expectations, level, possibilities and needs of their learners, thus constructing a ‘pedagogical grammar’ (PG) – “a functional grammar adapted for pedagogical purposes, in the form of practically-oriented rules, definitions and explanations supposed to facilitate the assimilation and generation of grammatical target structures in the most economical way.” This is the pivotal concept for this thesis and it therefore requires some elaboration, but prior to that let us just mention one more type of grammar to complete the paradigm. This is “the learner’s own ‘psycholinguistic’ grammar” (LG; Tonkyn 1994:1, emph. added) representing his/her competence in the FL (≈ ‘the learner’s IG’) reflected in his/her interlanguage, basing on what s/he has managed to internalise from the input and the PG presented to him/her. The whole progression can schematically be presented in the following way:
exposure ð Learner’s Grammar
Figure 1: Interrelatedness of four kinds of grammar
The diminishing fonts and arrows and the dashed horizontal lines are used to represent the consecutive filtering of the information and the ‘invisibility’ of some aspects of knowledge at each stage before it is eventually available to the learner. IG (in the form of corpus data, NSs’ intuitions, or both) and EG form a frame of reference, whereas in the process of teaching a language one mainly concentrates on the other two grammars. As depicted here, PG typically has no direct access to the internalised grammar of the NS (unless through exposure only); it can only be derived from EG, and should therefore be as descriptive as possible rather than prescriptive. It acts as a filter, containing a selection, adaptation from EG, adjusted to meet the needs of the learners.
The learner’s grammar will in most cases never achieve the level of the NS’s internalised grammar. In the incipient stages of learning LG has little in common with IG; it is mostly misconceptions that the learner has formed about the language system s/he is learning. Professor Romuald Gozdawa-Gołębiowski (p.c.) suggests that the relation between the two kinds of grammar can be illustrated as in Figure 2:
Figure 2: The relation between IG and LG in the early stages of IL development
The best that the learner can attain is developing a LG which fully fits within the idealised native speaker’s IG. The teacher’s primary goal, therefore, should be to help learners eradicate, and best prevent them from forming, false hypotheses about the FL (i.e., LG IG), simultaneously providing them with rules that will help them expand their LG to gain as much NSs’ competence as possible approximating to the model presented in Figure 3:
Figure 3: The desired position of LG with relation to IG
The designation ‘pedagogical grammar’ is thus generally employed to denote the explicit treatment of elements of the TL system as part of language teaching methodology. It logically embraces all aspects of language teaching that in some way seek to make sense of and systematise the TL for the learner – thus, not only what to teach, but also how to set about it (see also Long 1985). The contribution of PG to language teaching is that firstly, it acts as Occam’s razor simplifying ‘scientific’ (i.e. ‘formal’, ‘linguistic’) grammar—PG should be easily comprehensible and not overly sophisticated as reference grammars tend to be—and modifying it in accordance with the conditions and goals of learning, making it accessible to the L2-er. Secondly, it meets their relevant needs. In other words, PG is “a cover term for any learner- or teacher-oriented description or presentation of foreign language rule complexes with the aim of promoting and guiding learning processes in the acquisition of that language” (Dirven 1990). If learner-centred, it is obviously going to be more informal; if teacher-centred, it will be much broader, containing pedagogical implications and suggestions how to teach the structures. Corder (1973) suggests the following hierarchy in which pedagogical grammar is the lowest stage, which I present here schematically:
Presentation of language for teaching
Figure 4: A hierarchy in linguistic studies (based on Corder 1973:156)
As in the previous diagram, I use the shrinking fonts and arrows and dashed horizontal lines to represent the consecutive filtering of information at each stage in the process of adapting it for the needs of the learners.
It must once again be highlighted at this point that PGs are addressed to not only students, but teachers as well (if not predominantly). However, in this dissertation here and henceforth the term ‘pedagogical grammar’ will be used to refer not to a model which the teacher uses for reference, but to one intended for eventual presentation to the learner with the aim of making him/her familiar with the FL system and helping him internalise the rules.
source: Paradowski, Michał B. (2007) Exploring the L1/L2 Interface. A Study of Polish Advanced EFL Learners. Institute of English Studies, University of Warsaw, pp. 92–5.
 In reality, each speaker operates his/her own idiolectal IG, which may vary owing to regional, educational, community and other differences. Also, some writers are turning more and more to corpora of interactions of successful L2 users rather than NSs as the models. After all, the goal of FL education is not to produce linguistically deprived monolinguals in the predominantly multilingual population of today’s world: “[w]e are pretending when we tell our students that a monolingual environment filled with monolingual speakers is authentic, according to any real-life norms” (Chavez 2003:194). Although on the surface it is English that seems to be the goal of ELT, this actually means teaching the learners to be bilingual! Consequently, (not only in the domain of linguistic accuracy, but also in interlanguage pragmatics; House & Kasper 2000:101ff.) a more adequate and appropriate language model (or reference group) for EFL/ELF students—shifting the focus to the realities of global uses of English—would be that of successful, fluent (if not necessarily expert) bi-/multilingual L2 user rather than the woefully inadequate exemplar of a monolingual native speaker talking to a compatriot. Since his/her IL knowledge and skills are under construction, “the yardstick by which the unstable bilingual should be measured is the stable bilingual under comparable social, cultural and historical conditions of language use, and with comparable goals for interaction in different discourse domains” (op. cit.:111); thus, using language-testing terminology, the 100%-criterion-referenced perspective ought to give way to a NNS-norm-referenced one. While reference materials on the performance of NSs may serve an informative purpose for a linguist, they may no longer be so useful to the learner. Increasingly, rather than teachers’ obsolescent prescriptive norms of correctness, learners’ contexts of English-language usage are going to be the main reference point, and some ‘authentic’ NS materials should be replaces with ones based on corpora of successful NNS interactions. This is also recognised in the specification of CEF level descriptors, whose intention is evidently not to progress to the level of native-speaker competence:
Level C2 includes descriptions of the high level, complex language an educated, highly trained learner of a foreign language may be able to use, and it includes skills which a majority of native speakers do not possess. (Heyworth 2004:17; emph. added)
Naturally, the qualifications for teaching bilingualism necessarily mean bilingual teachers.
 The E-language of a speech community is often frequently defined as the aggregate output of the distinct I-languages (idiolects) of its members.
 As observed by Grzegorz Śpiewak (p.c.), EG is not really ‘smaller’ that IG; if at all, we would expect it to be bigger (cf. fn. 154). In corpus-informed grammars, such as the LGSWE, we will find information that is unavailable to some native speakers themselves (e.g. formal styles and registers). Even qualifying the term IG and restricting it to educated native speakers, it can reasonably be argued that the insights gained from corpus analysis supersede that of any single (non-linguist) speaker of a language. Thus, probably the relationship between EG and IG should better be characterised as partial overlap and difference in quality, if not necessarily in ‘size’.
 Obviously, we cannot reasonably expect a PG to be fully linguistically comprehensive and complete; this would be similar to expecting a learner to develop his/her linguistic competence “from zero to native or near-native competence in one leap” (Mohammed 1996:284). Thus, in a manner similar to IL, PG might actually better be conceived of as a continuum.
 The question might be posed which is more desirable: a smaller LG totally contained within the IG, or a larger one yet ‘spilling’ outside (Prof. Romuald Gozdawa-Gołębiowski, p.c.).
 The highest, initial state may be perceived as debatable, as it may carry strong connotations with axiomatic systems such as X-bar syntax, Minimalism, or HPSG. For the purpose of language pedagogy, and with impressive language corpora available to grammarians nowadays, this stage can practically be disposed of.