These two volumes contain 55 papers presented at the 36th Annual Boston University Conference on Language Development (BUCLD) held November 4-6, 2011. The annual BUCLD conference is organized by students in the Linguistics Program at Boston University, and attracts papers from leading researchers over the world as well as emerging new researchers, and represents a wide range of theoretical and methodological approaches to the study of language acquisition.
Due to the number of papers presented at this conference, only the keynote speaker's and a number other papers that were of particular interest are reviewed. Because of space limitations, little with respect to theoretical aspects is presented here. Interested readers are encouraged to read the original texts for a more in-depth understanding.
Excluding the plenary talk, published as the initial article in the collection, all chapters are organized alphabetically by author. The complete table of contents can be seen on the Cascadilla Press web site.
The first paper is the plenary address by Cornelia Hamann, entitled “Bilingual Development and Language Assessment” (pp. 1-28). This paper presents a review of work undertaken by Hamann and colleagues as well as other members of the COST (European Cooperation in Science and Technology) Action IS0804 (S. Armon-Lotem, PI, http://www.bi-sli.org/) on bilingualism and specific language impairment (SLI), and the role bilingualism plays in evaluation of language disorders. The author makes a principled distinction between different types of bilingualism from the onset (i.e., simultaneous / successive; within successive: adult / child; within child: early / late) and points out that bilingualism is the norm rather than the exception in the world and quite common in Europe, even in smaller cities. The author highlights the fact that SLI is over-diagnosed in bilingual children, and points out a few subtle differences that emerge between L2 and SLI profiles. A case study of two bilingual Russian-German children is presented, where details of one child's abilities and disabilities point to a diagnosis of SLI, while the second seems to show normal patterns of delay related to L2 learning that should resolve in time. However, the details of the evaluation are based on analyses of spontaneous speech and thorough syntactic analyses, which are not always favoured by speech language pathologists, due to their time-consuming nature (and probably also to the need to be comfortable basing one's syntactic analyses on target language vs. L1 influences). One answer to this problem is to develop more cost-effective approaches, which are ongoing in the COST Action IS0804. A preview is presented of research in progress in these domains. Parental questionnaires, tasks investigating morphology, syntax and computational semantics as well as working memory and executive function are being developed for different languages (and language pairs) by this team.
Of the other fifty-five chapters in the collection -- covering topics on theoretical and methodological approaches to research on language acquisition and cognitive development -- a number that grabbed my attention are described here.
Almeida et al. (pp. 42-52) present a paper entitled ''Prosodic Influence in Bilingual Phonological Development: Evidence from a Portuguese-French First Language Learner''. This research on a single speaker investigates early cross-linguistic influences on the acquisition of phonological segments and structures (i.e., codas and branching onsets) in a Portuguese-dominant simultaneous Portuguese-French bilingual. The data appear to support the acceleration of the acquisition of branching onsets in Portuguese due to the influence of French, concurrently with a delay in the acquisition of French word-medial codas due to the influence of Portuguese. These data question language dominance impacts on bilingual language acceleration / deceleration effects.
Armon-Lotem and Chiat (pp. 53-62) present data on L2 non-word repetition (NWR) in Hebrew and Russian. In order to test whether NWR is different in L1 and sequential L2 learners, whether it differs within the L1 and L2 of sequential L2 learners and whether word-likeness and phonological complexity affect NWR similarly in L1 and sequential L2 learners, they control for word length and morphological structure (derivational types) in the languages studied. They show that, contrary to what has been reported before, L1 and sequential L2 learners showed similar abilities on NWR, in that both groups showed more difficulties on long vs. short NWs, while long Hebrew NWs were more difficult than long Russian NWs for both groups. This last effect was explained by the fact that quadrisyllabic Hebrew NWs were created using a less frequent morphological process (abstract words in the pattern hitCaCCeCut, acquired after age 5) than Russian NWs of similar length (lad [stem] -av-och-k [derivational suffixes] -a [fem]) which are not abstract. This study showed the usefulness of using these tasks with polyglot children, if one can create comparable stimuli across the child's languages.
Arunachalam et al. (pp. 63-73) present a study of verb acquisition in interaction with the adverbs 'slowly' and 'nicely' in English children aged 2-2;5 using a preferential looking paradigm. They show that the manner of motion adverb 'slowly', but not manner adverb 'nicely', promotes verb learning in these children (contrary to previous results from the same group with similarly-aged children, but in that case no adverb was used, and no effects were found) even though children normally do not use this specific adverb at this age. Since only two adverbs were used, two explanations are suggested for the difference between results on 'slowly' vs. 'nicely': 1. Semantic properties of the stems 'slow' and 'nice' or 2. The higher frequency of 'slowly' as compared to 'nicely'. Future studies controlling these properties will be able to distinguish these two explanations.
In ''Performance Factors Trump Representational Deficits: Perception and Production of English Inflections by L1 Mandarin Speakers'', Bonner and Martohardjono (pp. 74-86) present a study on the perception and production of English inflection by L1 Mandarin speakers learning English as an L2. High and low intermediate speakers of English recruited in Beijing took part in experiments on the perception and production of past tense and plural English morphemes, in simple nouns and in sentences. Equal numbers of syllabic (Vd/Vz) and non-syllabic (t/d, s/z) contexts were created. In perceptual tasks no effects were found for 1. inflection type, 2. participant group, 3. syllabicity. Only one effect of voice was found (d > t). Within sentences, inflection type showed effects while participant group did not. Syllabic suffixes were better recognized than other types. Within non-syllabic structures opposite patterns for voice are found where d > t but s > z. Results on production tasks show slightly different performance where no effects were found on single word production and only one effect of inflection type was found where Vz < z. These data show that perceptual difficulties do not preclude accurate production of inflection (or vice versa) and that production difficulties in L2 learning are probably not due to representational deficits.
Cantiani et al. (pp. 114-125) present what I expected to be exciting data comparing off and on-line (event related potential, or ERP) data on inflexional morphology in children with developmental dyslexia (DD) and children with DD in addition to language impairment (DD+LI). This paper is discussed further in the Evaluation below.
Culbertson and colleagues (pp. 139-151) present a neat artificial language learning experiment where they manipulate pattern majority/minority status as well as two-word (Noun-Adj) and three-word (Det-N/Adj-N) orderings based on universal typologies that have been argued to be constrained by innate preferences. Generally the results confirm that language speakers prefer 'harmonic' typologies where both Adj and Det precede or follow the N, independently of the input statistics.
Fleischhauer and Clahsen (pp. 164-176) investigate verb form generation while controlling for age, frequency, and working memory in German-speaking children. They show that frequency effects are stronger on irregular verbs (low-frequency ones being produced more slowly than high-frequency ones), and that regularity and frequency show paradoxical interactions where irregulars are produced faster when they are frequent, while regulars are produced faster when they are lower frequency. This is almost impossible to explain from a connectionist perspective. This result is also specifically linked to STM scores in adults and all their child data, supporting their interpretation that the paradoxical effects are linked to competition between memory-based and decomposition processes competing in regular word production.
Grüter et al. (pp. 213-225) present eye-tracking data in a study of object clitics in Spanish children and adults. They also investigate whether children who omit clitics can still process them. Their data show that this is not in fact the case: 4-year old Spanish speakers who do not produce clitics do not use clitic information when anticipating object nouns in depicted scenes. These results were not expected on a production-only hypothesis, nor a representational deficit where output is expected to be variable but comprehension good.
Hopp (pp. 226-245) also presents eye-tracking data, but on gender and number integration in late L2-speakers of German. He shows that at advanced levels of L2 learning, these speakers can integrate gender and number information rapidly and incrementally, just as L1 speakers do, although they might show slight delays (in terms of milliseconds) in this ability. Furthermore, the division of participant groups into variable vs. consistent producers of correct gender agreement shows that only the second group uses gender marking predictively in eye-tracking.
Maguire et al. (pp. 328-338) study object-noun and action-verb identification using event related potentials (ERPs) in adults and children aged 8-9 using a picture-word matching paradigm. They present data they argue supports a feature-based distinction between objects and verbs, as an early signal related to semantic coherence (the N300) shows larger effects of congruence (visual/word match) for verbs than nouns in adults, and only effects for noun incongruency in children.
In ''The Acquisition of Distributivity in Pluralities'', Pagliarini et al. (pp. 387-399) present a study on the interpretation of definite plural noun phrases in close to 200 children aged 4 to 13 years old. They evaluate the two possible interpretations of sentences like 'each boy is building a snowman' and 'the girls are building a snowman', where there are one or two snowmen (under their hypotheses that children tolerate degraded distributive readings of definite plural NPs). Their data show that both sentences with 'each' and with definite plural NPs were accepted as having collective (e.g., each boy is building a single snowman) or distributive (e.g., each girl is building a different snowman) readings, by younger children. Only older children understand that the use of 'each' specifically implies distributional readings and thus pragmatically disallows it for definite plural NPs.
Parr and Breheny (pp. 427-436) show that the use of bare NPs in child language corpora correlates with the type of utterance the child is using. They find that bare NPs correlate with Manifest Events (MEs, descriptive, present objects, etc.) but not Complex Events (CEs, which can be INTentional or RESultative), which in turn are more highly correlated with complex DP production. The authors argue that the need of the child to communicate intentions clearly make the INT events particularly important for the development of complex DPs and VPs.
Pirvulescu et al., in ''Clitic Production across Tasks in Young French-Speaking Children'' (pp. 461-473), present data supporting the optional nature of accusative clitic production in French, by keeping context constant while varying output demands (production of clitics in the 3rd person singular and 2nd person singular present, or imperative mood). They show that task modulations influence the number of clitics a child will produce in elicitation, with the 2nd person singular condition being the most congenial one for clitic production.
Poepsel et al. (pp. 474-486) show in ''Context, Mutual Exclusivity, and the Challenge of Multiple Mappings in Word Learning'' that while multiple meaning mappings for the same word form can be difficult to learn, specific contexts promote their learning. In tasks with adults, they show that presenting different word-visual pairings with different voices, different voices and accents, or explicit instructions, promote the ability to learn multiple mapping in similar ways.
Tanner et al. (pp. 594-606) present an ERP study on L1 and late L2 learners of English (Spanish L1) checking noun-verb agreement processes with intervening 'attractor' nouns. They show the ERP waves are modulated by the type of structure in which the attractor noun is located (PP or Relative Clause) and that, although L2-speakers show weaker effects than L1-speakers, these interactions are the same in both groups. Ungrammatical structures elicit late positivities (P600 in both groups) that are stronger for singular attractors than plural ones and stronger for attractors in RC versus PP structures.
Vasić et al. (pp. 646-659) study gender acquisition in both L2-Greek and Dutch by L1-Turkish children using a self paced listening task. They show that target language morphological transparency impacts strongly on the child's ability to master and integrate the system.
Wen & Schwartz (pp. 673-685) argue in their paper that task demands can strongly influence our understanding of language processing in L2 speakers. In a structure-focused self-paced reading task they find that Chinese-L1 English-L2 speakers are able to process morphosyntactic agreement and are also sensitive to subject-verb agreement errors, contrary to what has been found in content-based (semantic verification) tasks.
Yu et al. (pp. 686-697) present ''Electrophysiological Correlates of Picture-Word Processing in Three-to-Seven Year Old Non-verbal Children with Autism'' using a picture-word (auditory) presentation paradigm. They show that non-verbal children with autism do not show typical negativities (N400) to visual-auditory mismatches, while typically developing children do. In addition, a sub-group of the children with autism who do have some vocabulary do show these N400s. All children showed early auditory (P1) visual peaks (P2), indicating that sensory disorders were probably not the cause underlying the absence of N400 effects.
Many more papers are presented in the two volumes, dealing with vast domains of inquiry ranging from theoretical aspects of language acquisition to computational modeling of language acquisition and processing. In addition, a large number of languages are studied including French, German, Greek, Italian, Japanese, Korean, Mandarin, Portuguese, Spanish, Turkish and others. Different learning profiles including early bilingualism and late L2-learning, dyslexia, SLI and sign languages are also addressed.
Linguists and psycholinguists, speech-language pathologists and others interested in the development of phonology, morphology, syntax and semantics, in monolingual, bilingual and language-disordered populations, will find a wide variety of research articles in these BUCLD 36 volumes.
The methodological approaches and theoretical assumptions are quite varied in the papers, thus making the papers extremely variable in their scope and coherence with the rest of the volume. They are written as ‘stand-alone’ papers, and thus the reading of a given paper does not oblige the reader to read others. The most interesting aspect of these is that they report on very recent research, which is often not yet available elsewhere (with a few exceptions which seem in fact to have already been published at least in part). Chapter quality is quite variable, with some research still ongoing, some methodologies questionable, or theoretical assumptions not explicit. In particular, statistical analyses can be quite intricate and well thought-out (using multiple regression analyses, for example), while others are extremely poor. Some analyses (whether on response, reaction-time or even ERP data) do not respect the basic tenet that if you do not find an interaction of effects, you cannot break down analyses into different partitions (Nieuwenhuis et al., 2011). Some papers do not even bother to check this assumption. For example, the ERP data by Cantiani et al. (pp. 114-125) are not appropriately analyzed. They discuss differences in ERP patterns between their three groups (children with no impairment, children with developmental dyslexia, DD, and children with DD in addition to language impairment, DD+LI). However, they never show any statistics on group differences or interactions in ERP waves that motivate their subsequent individual group analyses. Furthermore, the data are hard to interpret, as the experimental items used in off and on-line tasks are different (that is, novel plurals and novel word formation -- verb participles, and derivational processes--, versus grammaticality judgment of subject-verb number agreement, respectively). However, in general, the quality of the papers is quite high, with clearly presented theoretical assumptions, methodologies, analyses and results.
It is possible to link up themes in the papers, as hot topics tend to pop up throughout them. In this year's BUCLD proceedings, I noted the frequent use of cutting-edge dynamic techniques for the study of language (eye-tracking, self paced-reading, ERPs), quite detailed approaches to L2 learning (e.g., different L1 backgrounds, different levels of L2 attainment, acceleration vs. deceleration effects) and a strong interest in agreement processes. Other readers will make links between other topics that are of specific interest to them.
The editing work on these volumes is better than it used to be: there is much less variability between papers in terms of language quality, typos, and reference style. However, I have previously commented on the fact that too many citations are to conference presentations that are not available in print (http://linguistlist.org/issues/19/19-750.html). The editors should require authors to only cite articles or manuscripts that are readily available (on personal websites, for example), or ask the authors to make them available if they wish to cite them.
The papers in these volumes are directed at researchers and graduate students in language acquisition and language learning. Because of the short length of the articles, a strong background is necessary to be able to appreciate their contents. However, undergraduates could also benefit from these papers, especially if put in the context of other readings providing more context for the understanding of theoretical and methodological issues. Their short length also allows these articles to be used as discussion papers in seminars and for undergraduate courses.
Nieuwenhuis, Sander, Birte U Forstmann & Eric-Jan Wagenmakers. 2011. ''Erroneous analyses of interactions in neuroscience: a problem of significance.'' Nature Neuroscience 14: 1105-07.
ABOUT THE REVIEWER:
Phaedra Royle holds a Ph.D. in Linguistics from the Université de Montréal and pursued postdoctoral studies at the School of Communication Sciences and Disorders at McGill University. Her interests lie in psycholinguistics, neurolinguistics, language disorders, language acquisition, morphology, morpho-phonology and morpho-syntax, and processing of complex noun phrases in French populations with and without learning challenges (SLI, Cochlear implants, Bilingualism, Ageing). She is a professor at the School of Speech Language Pathology and Audiology at the Université de Montréal, and is a member of the Centre for Research on Brain, Language and Music.