Goldrick, Putnam and Schwarz (Goldrick, Putnam & Schwarz) have offered an account that must delight all linguists who have spent an inordinate amount of their professional time working on understanding the grammar of bilingual language use – just how, and why, do bilinguals produce the utterances that they do. The computational approach Goldrick et al. propose blends the grammatical principles (of the optimality kind) with general processing constraints to yield patterns of code-mixing that may be sparse – the emergence of doubled elements in bilinguals’ utterances – but certainly needing an explanation. Their Gradient Symbolic Computation model is, in fact, the first robust account of the presence of doubled elements, i.e., an element of the utterance is doubled, appearing in both languages within a single utterance (see 1 below). Their account, however, does force them to make an important, but vulnerable, theoretical assumption: that grammatical principles can refer to language membership. It recalls, albeit obliquely, Belazi, Rubin and Toribio's much discussed ‘Language’ feature needed to mobilize their Functional Head Constraint (see Bhatt, Reference Bhatt1997, for an empirical and theoretical critique of the Functional Head Constraint).
Goldrick et al. present their account of code-mixing in terms of three general principles; (i) a principle that uses weighted constraints to capture cross-linguistic as well as intra-speaker variation, (ii) a principle that uses language-specific weighting constraints and (iii) a principle that assumes a simultaneous coactivation of representational elements in both the input and the output of the grammar, allowing for representations that blend elements from multiple languages. Although their second principle associates each language with distinct weighting of constraints, it is not clear what precise, principled mechanisms of weight adjustments on constraints will yield the inter-speaker variation, given that, as they claim, bilingual speakers have varying degrees of competence in multiple grammars. In other words, since the command of two (or more) languages varies from speaker to speaker – from ambilingualism to incipient bilingualism – in a given multilingual community, how are the weights of different constraints adjusted to reflect the nature of individual bilingual competence. It is not clear to me how their Principle 2 (Gradient blends of grammar) offers a clear explication of this issue of inter-speaker variation.
These minor quibbles notwithstanding, I applaud Goldrick et al. for providing a thorough discussion of blend structures, for the English–Tamil doubling construction they gave me a grant kodutaa. That discussion is clear and precise. One, of course, wonders if a similar theoretical toolkit will be recruited to account for a construction-type closely related to doubling, known as the portmanteau constructions, which is often reported in the studies of bilingual utterance (e.g., Nishimura, Reference Nishimura1985; Azuma, Reference Azuma1993). A ‘portmanteau sentence’ is one that has a hybrid structure from two sentences in different languages. In this type of sentence, a constituent in one language is shared as a constituent in another language (Azuma, Reference Azuma1993: 199). The following Japanese–English code-mixed examples show such construction type:
-
(1)
In (1) above, each underlined part is shared by the two languages, and the word that immediately follows the shared constituent corresponds categorically and/or semantically to the one that immediately precedes it. In (1a), for instance, the English object ‘two pounds’ is shared as a constituent in both English and Japanese, resulting in the symmetrical configuration of (S)VOV. The schematic illustrations of (1a) are given in (2) and (3) below:
-
(2)
-
(3)
This mirror-image structure, in (3) above, is possible due to the opposite word order in both languages (i.e., English, an SVO language, and Japanese, an SOV language) as well as the ellipsis of subject in Japanese, which is very common in informal speech. However, this construction is normally inhibited due to the fact that it violates the well-established syntactic condition in (4) below:
-
(4) The Extension Condition
-
A syntactic derivation can only be continued by applying operations to the root projection of the tree.
Conversely, the Extension Condition disables syntactic operations from occurring at the level of terminal nodes. As is shown in the diagram in (3), however, the Japanese postposition P2 ‘gurai’ merges with the terminal node of the English prepositional phrase P1P, i.e., the shared constituent DP ‘two pounds’, which is already merged with the English preposition P1 ‘about’. The question that needs to be addressed now, using the Gradient Symbolic Computation model, is: How do bilinguals generate portmanteau sentences despite the fact that the Extension Condition invalidates the derivation of them?
Goldrick, Putnam and Schwarz (Goldrick, Putnam & Schwarz) have offered an account that must delight all linguists who have spent an inordinate amount of their professional time working on understanding the grammar of bilingual language use – just how, and why, do bilinguals produce the utterances that they do. The computational approach Goldrick et al. propose blends the grammatical principles (of the optimality kind) with general processing constraints to yield patterns of code-mixing that may be sparse – the emergence of doubled elements in bilinguals’ utterances – but certainly needing an explanation. Their Gradient Symbolic Computation model is, in fact, the first robust account of the presence of doubled elements, i.e., an element of the utterance is doubled, appearing in both languages within a single utterance (see 1 below). Their account, however, does force them to make an important, but vulnerable, theoretical assumption: that grammatical principles can refer to language membership. It recalls, albeit obliquely, Belazi, Rubin and Toribio's much discussed ‘Language’ feature needed to mobilize their Functional Head Constraint (see Bhatt, Reference Bhatt1997, for an empirical and theoretical critique of the Functional Head Constraint).
Goldrick et al. present their account of code-mixing in terms of three general principles; (i) a principle that uses weighted constraints to capture cross-linguistic as well as intra-speaker variation, (ii) a principle that uses language-specific weighting constraints and (iii) a principle that assumes a simultaneous coactivation of representational elements in both the input and the output of the grammar, allowing for representations that blend elements from multiple languages. Although their second principle associates each language with distinct weighting of constraints, it is not clear what precise, principled mechanisms of weight adjustments on constraints will yield the inter-speaker variation, given that, as they claim, bilingual speakers have varying degrees of competence in multiple grammars. In other words, since the command of two (or more) languages varies from speaker to speaker – from ambilingualism to incipient bilingualism – in a given multilingual community, how are the weights of different constraints adjusted to reflect the nature of individual bilingual competence. It is not clear to me how their Principle 2 (Gradient blends of grammar) offers a clear explication of this issue of inter-speaker variation.
These minor quibbles notwithstanding, I applaud Goldrick et al. for providing a thorough discussion of blend structures, for the English–Tamil doubling construction they gave me a grant kodutaa. That discussion is clear and precise. One, of course, wonders if a similar theoretical toolkit will be recruited to account for a construction-type closely related to doubling, known as the portmanteau constructions, which is often reported in the studies of bilingual utterance (e.g., Nishimura, Reference Nishimura1985; Azuma, Reference Azuma1993). A ‘portmanteau sentence’ is one that has a hybrid structure from two sentences in different languages. In this type of sentence, a constituent in one language is shared as a constituent in another language (Azuma, Reference Azuma1993: 199). The following Japanese–English code-mixed examples show such construction type:
(1)
In (1) above, each underlined part is shared by the two languages, and the word that immediately follows the shared constituent corresponds categorically and/or semantically to the one that immediately precedes it. In (1a), for instance, the English object ‘two pounds’ is shared as a constituent in both English and Japanese, resulting in the symmetrical configuration of (S)VOV. The schematic illustrations of (1a) are given in (2) and (3) below:
(2)
(3)
This mirror-image structure, in (3) above, is possible due to the opposite word order in both languages (i.e., English, an SVO language, and Japanese, an SOV language) as well as the ellipsis of subject in Japanese, which is very common in informal speech. However, this construction is normally inhibited due to the fact that it violates the well-established syntactic condition in (4) below:
(4) The Extension Condition
A syntactic derivation can only be continued by applying operations to the root projection of the tree.
Conversely, the Extension Condition disables syntactic operations from occurring at the level of terminal nodes. As is shown in the diagram in (3), however, the Japanese postposition P2 ‘gurai’ merges with the terminal node of the English prepositional phrase P1P, i.e., the shared constituent DP ‘two pounds’, which is already merged with the English preposition P1 ‘about’. The question that needs to be addressed now, using the Gradient Symbolic Computation model, is: How do bilinguals generate portmanteau sentences despite the fact that the Extension Condition invalidates the derivation of them?