Lexical-phonological interactions in bilingual children.

33
First Language 2015, Vol. 35(2) 93–125 © The Author(s) 2015 Reprints and permissions: sagepub.co.uk/journalsPermissions.nav DOI: 10.1177/0142723715574398 fla.sagepub.com FIRST LANGUAGE Lexical-phonological interactions in bilingual children Margaret M. Kehoe Université de Genève, Switzerland Abstract This study examined lexical-phonological interactions in the first 50 words of a group of monolingual German- and Spanish-speaking children and bilingual German–Spanish children. The phonological characteristics of the earliest target word forms and output patterns of these children were analyzed to determine whether bilingual children select different target word forms and produce different output forms and templates than monolingual children. Results indicated that the target word forms selected by Spanish- speaking children differed from German-speaking children but not the target word forms of monolingual versus bilingual children. Output forms differed between monolingual German and Spanish children but did not differ greatly between the two languages of the bilingual children. The output patterns of bilingual children also differed from those of monolingual children. Bilingual children’s earliest words contained more /l/s and variegated place and their templates were characterized by more mature phonological forms. These patterns were interpreted as being due to cross-linguistic interaction or to a more general effect of bilingualism, which leads to increased phonetic-phonological sophistication. Keywords Bilingualism, first word period, lexical-phonological interactions, lexicon, phonological development, phonology Introduction Researchers have long been aware that phonology and the lexicon interact in the early stages of language development (see Stoel-Gammon, 2011, for a review). Studies show that phonological ability influences lexical acquisition, and that the structure and nature Corresponding author: Margaret M. Kehoe, Faculté de Psychologie et des Sciences de l’Éducation, Université de Genève, 42, bd du Pont-d’Arve, 1205 Genève, Switzerland. Email: [email protected] 574398FLA 0 0 10.1177/0142723715574398First LanguageKehoe research-article 2015 Article

Transcript of Lexical-phonological interactions in bilingual children.

First Language2015, Vol. 35(2) 93 –125

© The Author(s) 2015Reprints and permissions:

sagepub.co.uk/journalsPermissions.navDOI: 10.1177/0142723715574398

fla.sagepub.com

FIRSTLANGUAGE

Lexical-phonological interactions in bilingual children

Margaret M. KehoeUniversité de Genève, Switzerland

AbstractThis study examined lexical-phonological interactions in the first 50 words of a group of monolingual German- and Spanish-speaking children and bilingual German–Spanish children. The phonological characteristics of the earliest target word forms and output patterns of these children were analyzed to determine whether bilingual children select different target word forms and produce different output forms and templates than monolingual children. Results indicated that the target word forms selected by Spanish-speaking children differed from German-speaking children but not the target word forms of monolingual versus bilingual children. Output forms differed between monolingual German and Spanish children but did not differ greatly between the two languages of the bilingual children. The output patterns of bilingual children also differed from those of monolingual children. Bilingual children’s earliest words contained more /l/s and variegated place and their templates were characterized by more mature phonological forms. These patterns were interpreted as being due to cross-linguistic interaction or to a more general effect of bilingualism, which leads to increased phonetic-phonological sophistication.

KeywordsBilingualism, first word period, lexical-phonological interactions, lexicon, phonological development, phonology

Introduction

Researchers have long been aware that phonology and the lexicon interact in the early stages of language development (see Stoel-Gammon, 2011, for a review). Studies show that phonological ability influences lexical acquisition, and that the structure and nature

Corresponding author:Margaret M. Kehoe, Faculté de Psychologie et des Sciences de l’Éducation, Université de Genève, 42, bd du Pont-d’Arve, 1205 Genève, Switzerland. Email: [email protected]

574398 FLA0010.1177/0142723715574398First LanguageKehoeresearch-article2015

Article

94 First Language 35(2)

of the lexicon influence phonological knowledge. Thus, there appears to be a bidirec-tional relationship between phonological and lexical development. This study examines this relationship in bilingual children. It focuses on lexical-phonological interactions during the early meaningful speech period, at a time when children’s productive capaci-ties play an important role in the selection of words and when children’s phonologies may be characterized by whole-word patterns or templates. Specifically, it considers lexical selection, output patterns (and templates) in the first 50 words produced by four German–Spanish bilingual children during naturalistic longitudinal recordings. The tar-get word forms and production (output) patterns of these children are compared to those of four German-speaking and four Spanish-speaking monolingual children in order to determine whether they are similar to or different from them.

Lexical selection

A variety of observational studies support the fact that some children display preferences for words with particular sounds or for sounds in specific syllable or word positions (Ferguson & Farwell, 1975; Menn, 1976). For example, Stoel-Gammon and Cooper (1984) reported that one of the three subjects in their study, Daniel, produced a high proportion of words which ended in a velar stop (e.g., quack, rock, clock, sock, whack, milk, frog, yuk, block, walk). Of Daniel’s first 50 words, 22% contained a final velar in comparison to 8% and 4% in the vocabularies of the other two children in the study. Important to note is that Daniel displayed a clear preference for consonant vowel conso-nant (CVC) structure and velar stops in his earliest lexical productions.

As many authors have noted, observational studies cannot exclude the possibility that chance factors play a role in lexical selection. Schwartz and Leonard (1982), in a land-mark study, provided experimental evidence to confirm the influence of phonological factors in lexical acquisition. They found that children aged 1;2–1;10 learned to produce nonsense words containing sounds that they could produce (i.e., phonologically ‘in’ words) more easily than nonsense words containing sounds that they could not produce (i.e., phonologically ‘out’ words), thus, showing that children’s phonological abilities influence the content of their early vocabularies.

Whole-word templates

Another way in which lexical selection may operate is through the use of ‘whole-word patterns’ or ‘templates’. One of the earliest reported examples in the literature was Waterson’s (1971) son P, aged 1;6, who produced words such as another, finger, Randall, and window with a similar output pattern, namely a CVCV form in which both conso-nants were nasals (e.g., [ɲaɲa], [ɲeːɲeː], [ɲɪːɲɪ], [ɲaɲø], [ɲeːɲeː]). P appeared to focus on certain salient features of the target words (e.g., the presence of nasal consonants) and then reproduce these features using his own established articulatory patterns.

Since Waterson’s (1971) study, numerous researchers have observed the presence of whole-word patterns in children’s speech (Ferguson & Farwell, 1975; Macken, 1979; Menn, 1983); however, only in recent years has a more elaborated approach to whole-word phonology been developed in the form of Radical Templatic Phonology by Vihman

Kehoe 95

and colleagues (Velleman & Vihman, 2002; Vihman, 2002, 2009, 2010; Vihman & Croft, 2007; Vihman & Keren-Portnoy, 2013; Vihman & Velleman, 2000). According to Vihman, templates are a compromise between the phonological model provided by the target word and the child’s own phonetic repertoire of sounds and syllables (Vihman & Croft, 2007). Thus, templates stem from two sources: (1) implicit knowledge of the ambient language and (2) the child’s individual motor preferences. These templates lead to two types of processes in children’s speech: Selection and Adaptation. Children select words that con-form to their template and they adapt words to conform to their template. At the begin-ning, words are typically selected only and, therefore, they often appear as accurate matches to the adult target. Later, children project their own production routines onto adult words (i.e., adaptation) and they produce less accurate renditions of the target word.

For the purposes of our study on bilingual language development, it should be noted that cross-linguistic differences in templates have been reported. For example, Vihman and Croft (2007) report that a no-onset template (e.g., VCV) is characteristic of the speech of French-, Finnish-, and Welsh-speaking children but not of English-speaking children. One aspect of this study is to examine whether there are characteristic tem-plates in the speech of German- and Spanish-speaking children and whether they are also seen in the bilingual children’s speech.

One potential difficulty in studying templates is that we do not know how frequent they are in children’s speech. Indeed, whole-word patterns or templates have been fre-quently discussed in the literature; yet, there have been no large-scale studies of template use. The same applies to lexical selection. We know that some children display patterns of lexical selection, but we do not know whether all children do. Another difficulty is that there is no operational definition of a template. Template descriptions may refer to a syl-lable or word form pattern (e.g., CVCV disyllables of Italian children, Keren-Portnoy, Majorano, & Vihman, 2009), a melodic pattern (e.g., consonant harmony in Italian chil-dren, Keren-Portnoy et al., 2009), or a combination of both (e.g., Raivo’s final sibilant pattern <(C)Vs> or Tom’s medial-/l/ pattern <(C)VlV>, Vihman, 2002). Thus, descrip-tions of templates may range from general to specific.

Lexical selection and templates in bilingual children

Celce-Murcia (1978) reports on patterns consistent with lexical selection in an English–French bilingual 2-year-old. The child chose either the French or English equivalent of a word, based on her avoidance of phonologically difficult features. She chose couteau rather than knife but boy rather than garçon because her phonological system disfavored fricatives. Yavas (1995) described a similar strategy in his analysis of the first 50 words of Deniz, a bilingual Turkish–Portuguese child. Deniz’s avoidance of initial fricatives led to selection of Turkish /mum/ not Portuguese /vela/ ‘candle’, but Portuguese /kabelu/ rather than Turkish /satʃ/ ‘hair’.

With respect to templates, Vihman (2002) presents data from three bilingual children which reveal similar whole-word templates and word selection patterns across languages. For example, Raivo, an Estonian–English bilingual child, selected target words with sibilants and matched them to his own production pattern, which consisted of a CVC form with final ‘s’. He did this regardless of whether the source language was Estonian

96 First Language 35(2)

(e.g., [kys] küpsis ‘cookie’) or English (e.g., [dɪs] this). Tom, a French–English bilingual child, made use of a medial-/l/ pattern in both his French and English productions (e.g., French: pantalon ‘trousers’ [bəˈlo]; English: cardie (cardigan) [kaˈli]) even though medial-/l/ is not a frequently used pattern in English-speaking children.

The studies cited above suggest that bilingual children operate with a single phono-logical strategy that influences lexical selection and templatic patterns similarly across the two languages. For example, they apply a single strategy ‘select final-sibilants’ resulting in a penchant for words with final sibilants in both languages. Vihman (2009) argues that templates are mediated by the child’s own production. Therefore, articulatory practice and the fact that the child’s output serves as its own input play a role in why templates may be similar across the two languages.

Yet there is ample research evidence indicating that bilingual children develop separate phonological systems early in acquisition (Ingram, 1981; Keshavarz & Ingram, 2002; Lleó & Kehoe, 2002; Paradis, 2000). Therefore, it is conceivable that they develop differ-ent phonological strategies in each language, which in turn influence lexical selection and templatic patterns separately across the two languages. For example, a child may display a preference for final sibilants in one language but for final velars in the other language. Thus, bilingual children may not appear very different from their monolingual counter-parts in terms of lexical selection and templatic patterns. Nevertheless, there is also evi-dence for cross-linguistic interaction in phonological development (Fabiano-Smith & Goldstein, 2010; Kehoe, 2002; Lleó, 2006; Lleó, Kuchenbrandt, Kehoe, & Trujillo, 2003; Paradis, 2000). Bilingual children’s different phonological systems may influence their early selection of words in ways that are different from monolingual children’s. An exam-ple of this type of interaction could be the earlier selection and/or production of multisyl-labic words in German by German–Spanish bilingual children due to the influence of Spanish; multisyllabic words being more frequent in Spanish-speaking children’s early lexicons than German-speaking children’s (Lleó & Demuth, 1999).

At this stage, we cannot distinguish between these differing possibilities (one phono-logical strategy, two distinct phonological strategies, two phonological strategies but with interaction) because there have been few studies of word selection and templates in bilinguals. This is the aim of the current study. That is, we would like to determine if there is interaction not only between the phonological and lexical systems within a lan-guage but also between languages in the bilingual child. Figure 1 graphically summa-rizes the three different possibilities.1

Research analyses and questions

Lexical selection: A study of target word forms and output forms. To determine whether bilingual children select different words than monolingual children, we investigated the phonological characteristics of the first 50 target word forms of bilingual children and compared them to those of monolingual children. In this section, we consider which phonological characteristics to study in an analysis of children’s earliest words.

The earliest stages of phonological acquisition are characterized by unmarked phono-logical forms (Jakobson, 1941/1968). Children produce monosyllabic or disyllabic words containing CV syllables. In terms of manner and place of articulation, children produce

Kehoe 97

Figure 1. A schematic representation of possible lexical-phonological interactions in bilingual children.

Language A Language B

LexicalDevelopment

Lexical Development

Phonological strategy

Possibility 1. One single phonological strategy influences lexical selection similarly across languages.

Language A Language B

Lexical Development

Lexical Development

Phonologicalstrategy A

Phonological strategy B

Possibility 2. Different phonological strategies influence lexical selection differently across languages.

Language A Language B

Lexical Development

Lexical Development

Phonologicalstrategy A

Phonological strategy B

Possibility 3. Different phonological strategies influence lexical selection differently across languages but there is interaction between languages.

98 First Language 35(2)

plosives, nasals, and glides (few fricatives and liquids), and labials and coronals (few velars) (Stoel-Gammon & Dunn, 1985). Children also display consonant harmony pro-cesses or display restricted place-of-articulation (PoA) distribution, (Fikkert & Levelt, 2008; Keren-Portnoy et al., 2009; Vihman, 1978). Consequently, we coded target word forms in terms of those phonological characteristics that might be considered less typical of the first 50 word period and, thus, more likely to display cross-linguistic differences.

In terms of prosodic variables, we coded the percentage of multisyllabic words (3+syl-lables) and words with word-final codas. Spanish contains longer words (Lleó, 2006; Lleó & Demuth, 1999) and less complex syllable structure than German (Delattre, 1966). Thus, it could be predicted that Spanish-speaking children’s early lexicons contain more multisyllabic words and fewer words with codas than German-speaking children’s (Lleó, 2006; Lleó & Demuth, 1999).

In terms of manner of articulation, we code two features: sibilants and laterals. We code sibilants rather than fricatives since a sibilant pattern seems to be a preferred template during the early word period. For example, this has been observed in the English-speaking child studied by Waterson (1971) and in one of the three bilingual children studied by Vihman (2002). We code laterals rather than liquids, because rhotics are infrequent in early phono-logical acquisition whereas laterals are subject to cross-linguistic differences: /l/ is acquired late in English but emerges early in Spanish (Cataño, Barlow, & Moyna, 2009; Lleó, 2008). Thus, we make the prediction that laterals will be more frequent in target word forms in Spanish compared to German. In terms of PoA, we code %velars and PoA variegation, that is, the presence of word forms that contain two or more different supra-glottal places of articulation (i.e., 2+PoA). Lleó, Prinz, El Mogharbel, and Maldonado (1996) report that velars are more frequent in Spanish than in German, citing the findings of Ortmann (1975) for German, and Quilis and Esgueva (1980) for Spanish. Thus, we predict higher frequency of velars in Spanish compared to German. We make no predictions concerning cross- linguistic differences in %sibilants and %2+PoA due to lack of pertinent research.

To determine whether the patterns observed in word selection are also observed in word production, we examine the output patterns in terms of the same prosodic and seg-mental features coded in the target word form study.

Whole-word patterns: A study of templates. In addition to the quantitative analysis of output patterns, we examine the data in a more impressionistic fashion for the presence of whole-word templates similar to those described by Vihman and colleagues (Keren-Portnoy et al., 2009; Vihman, 2002; Vihman & Croft, 2007). In the absence of concrete guidelines on the frequency and phonological composition of templates, we developed our own guidelines which, we believe, are in the spirit of Vihman and colleagues’ use of templates.

1. Templates should be frequent. We used the criterion of 10% presence in the out-put forms of the first 50 words. The value of 10% should exclude any isolated forms which are not true templates but allow for the fact that not all templates may be sampled in the recording sessions. Thus, we did not use a stricter criterion of 20%.

2. Templates should be represented in both selection and adaptation processes. For the most part, we judged a phonological characteristic as a ‘template’ only when

Kehoe 99

a child selected words that matched the template and adapted words to this tem-plate. Both processes are inherent in Vihman and colleagues’ description of tem-plates and thus we believe should be included as a defining characteristic. On occasion, some salient patterns were present only on the basis of selection (or of adaptation) but were still judged to be templates. These exceptions are noted in the text.

3. Templates should be defined as narrowly as possible. Some descriptions of tem-plates refer to general syllable structure (e.g., CV, CVC), word form (e.g., CVCV) or melodic patterns (e.g., consonant harmony). In our view, these patterns are universal characteristics of early child language and are too all-encompassing to be designated as templates. We have endeavored to be as specific as possible in the identification of templates, defining them in terms of both syllabic and melodic patterns (e.g., CVClab, tVtV, VjV) wherever possible.

Research questions

The analyses of target word forms, output forms, and whole-word templates2 address the following research questions:

1. Do the phonological characteristics of target word forms, output forms, and whole-word templates in German- and Spanish-speaking monolingual children differ?

2. Do the phonological characteristics of target word forms, output forms, and whole-word templates in the German and Spanish of the bilingual children differ?

3. Do bilingual children display different patterns from monolingual children with respect to the phonological characteristics of target word forms, output forms, and whole-word templates in their respective languages?

Method

The data stem from a bilingual project in which children acquiring German and Spanish in Hamburg were followed longitudinally from the onset of word production through to 3;0 years. Monolingual data were taken from a previously conducted project, in which German children in Hamburg and Spanish children in Madrid were recorded longitudi-nally in similar data collection conditions to the bilingual project. In all cases, children were audio- and, in some cases, video-recorded in their homes while playing and inter-acting with one parent and one experimenter. The monolingual children were recorded using a high fidelity cassette recorder (Sony TC-D10 PRO) and directional microphone (Beyerdynamic). The bilingual children were recorded using a mini-disc recorder (Sony MZ-R55CH) and directional microphone (Sony ECMT S120). In both monolingual and bilingual testing situations, the microphone was concealed in a vest worn by the child. (See Lleó et al., 1996, for additional information on the monolingual data.)

The bilingual children were visited by two separate teams: a German- and a Spanish-speaking team. If one of the parents was present, he/she had to be a native speaker of the language in which the recording session was taking place. Following testing, all sessions were glossed and phonetically transcribed by native speakers.

100 First Language 35(2)

Participants

The participants in this study include four monolingual German (Britta, Bernd, Marion, and Thomas), four monolingual Spanish (María, Juan, Miguel, and José), and four bilin-gual German–Spanish children (Simon, Nils, Jens, and Manuel). All children, monolin-gual and bilingual, were from middle-class educated families. All parents (mothers and fathers) had graduated from high school and had done university or tertiary-level studies. This was the case for monolingual and bilingual groups alike.

The bilingual children were children of Spanish-speaking mothers and German-speaking fathers, whereby each parent followed the ‘une personne, une langue’ rule by addressing the child in his/her respective language. The parents’ language of communi-cation was primarily German in the case of Jens, Manuel, and Nils, and both Spanish and German in the case of Simon. The main care person during the first three years of life for the bilingual children was the mother, who was the main provider of the Spanish input.

Procedure

Selection of target word forms and output forms. For the target word form analysis, we selected all target word forms documented in the recording sessions starting from the onset of speech until the 25-word session, that is, the session in which children spontane-ously produced at least 25 different identifiable words (Vihman, Macken, Miller, Sim-mons, & Miller, 1985). This resulted in approximately 50 cumulative target word forms. Target word forms included mama and papa, proper names, and consistent onomato-poeic forms (ticktock, wauwau). They did not include the words ‘no’ and ‘yes’, imitated forms, nor variable onomatopoeic forms (e.g., meow, which was subject to different phonetic renditions). Table 1 displays the number of target word forms and the age period from which they were taken.

In the monolingual data, the 25-word session criterion occasionally resulted in the selection of fewer than 50 target word forms. For example, Bernd produced 22 target word forms at 1;8.06, 7 target word forms at 1;8.21, and then 72 target word forms at the next recording session, which took place more than one month later due to scheduling difficulties. We thus decided to take his cumulative target word forms up until 1;8.21, so as to avoid including words beyond the first word stage. This resulted in only 31 target word forms.

The bilingual data posed some additional challenges. As Table 1 shows, the bilingual children produced their first words in the recording sessions two to three months later than the monolingual children. All of the bilingual children (particularly Simon and Jens) reached the 25-word session earlier in Spanish than in German. For us to have main-tained the criterion of the 25-word session (or approximately 50 target word forms) in German, we would have had to have included words from later sessions (after 1;10 or 1;11), at a stage in which phonological and grammatical structures seemed to suggest that they were beyond the first word period.3 We, thus, decided to include fewer target word forms in German.

The selection of output patterns was less straightforward than the selection of target word forms due to the fact that a single target word form could be subject to several

Kehoe 101

different output forms. We follow the recommendations of Ingram (2002), Keren-Portnoy et al. (2009), and Majorano and D’Odorico (2011) in selecting a single output form for each target word form. If there was more than one output form, the most frequent form was selected. If there was no dominant form, the last one produced was selected.

Coding of target word forms and output forms. Children’s target word forms and output forms were coded according to the six phonological characteristics mentioned above. In terms of prosodic characteristics, the percentage of target word forms or output forms containing three or more syllables (%3+syllables)4 and word-final coda consonants (%codas) was calculated for each monolingual and bilingual child across the two lan-guages. Word-final coda consonants were counted as they are generally acquired earlier than word-medial codas and may be present in children’s earliest words (Bernhardt & Stoel-Gammon, 1996; Grijzenhout & Joppen, 1999).

In terms of segmental characteristics, the percentage of target word forms or output forms containing one or more sibilants (i.e., /s, z, ʃ, ʒ, ç, ʦ, ʧ, ʤ/) (%sibilants), one or more lateral consonants (i.e., /l/) (%laterals), one or more velar consonants (i.e., /k, g/) (%velars), and two or more supra-glottal places of articulation was calculated (%2+PoA). Input forms purportedly containing the lateral palatal approximant /ʎ/ (e.g., caballo, llama) were not coded as lateral, since /ʎ/ is largely produced as the palatal approximant [j] in European Spanish. Input forms, in which velar stops become fricatives (e.g., [x, ɣ]) in intervocalic position, were coded as velars however. Output forms containing the

Table 1. Number of target word forms and age span for cumulative 50 target word forms.

German Spanish

Children No. of word forms

Age span No. of word forms

Age span

Monolingual German

Britta 53 0;11.13– 1;05.13 Marion 49 1;01.07– 1;06.07 Thomas 47 1;00.00– 1:07.15 Bernd 31 1;00.18– 1;08.21

Monolingual Spanish José 40 1;02.11– 1;06.22Miguel 44 1;01.03– 1;06.07María 61 1;01.01– 1;06.03Juan 50 1;00.25– 1;08.05

Bilingual Nils 44 1;03.04– 1;08.05 51 1;04.21– 1;08.05Simon 29 1;03.04– 1;10.23 49 1;04.09– 1;09.06Jens 23 1;03.14– 1;11.14 56 1;02.07– 1;10.28Manuel 42 1;03.08– 1;07.16 53 1;03.08– 1;07.08

102 First Language 35(2)

lateral palatal approximant and velar fricatives were coded as lateral and velar respec-tively; these forms were infrequent in the data.

In the coding of PoA, inter-dental, alveolar, alveolar-palatal, and palatal consonants were all coded as a single PoA, namely, coronal. To provide an example of the coding of PoA, the Spanish word grande was coded as having 2+PoA, the [g] being dorsal and the [ɾ, n, d] being coronal; but the Spanish word cinta was coded as not having 2+PoA since the inter-dental fricative, alveolar nasal, and alveolar stop were all counted as having coronal PoA. The productions of grande as [dadɛ] and cinta as [hita] were both coded as not having 2+PoA, the segment [h] not having supra-glottal place.

Coding of whole-word templates. The production patterns of the first 50 target word forms were examined for the presence of whole-word templates. In the case of those children whose numbers of target word forms were considerably lower than 50 words (monolin-gual German Bernd and bilingual German Jens and Simon), we also included target word forms from the session following the 25-word point. Vihman (2002) points out that induction of a phonological pattern may be found some time after first word production, that is, in the first 50–100 target word forms. Thus, we believe that the inclusion of more target word forms in the case of these three children allowed for a more complete sam-pling of templates. The quantitative analysis of target word forms and output forms for these three children, however, was based on data up to the 25-word point only. Identifica-tion of templates was based on the guidelines mentioned above. Any exceptions to these guidelines are noted in the text. Please note that a given output form could be consistent with more than one templatic pattern. For example, Bernd’s production of bitte as [pipipɛ] was counted as part of both a consonant harmony and CVCVCV template.

Reliability of phonetic transcriptions. A sample of the monolingual and bilingual databases (approximately 12%) was subject to reliability tests. Agreement between different tran-scribers (native Spanish speakers for the Spanish data and German speakers for the Ger-man data) was 89% (Spanish) and 88% (German) for CV structure (i.e., agreement as to whether a segment was deleted or inserted); 84% (Spanish) and 90% (German) for agree-ment concerning manner of articulation features; and 90% (Spanish) and 97% (German) for agreement concerning PoA features. All of these results constitute acceptable reliabil-ity measures. The slightly lower results for Spanish in the agreement of manner features was due to confusion between plosives versus fricatives (e.g., transcriber 1: [ˈβeɪβo] vs. transcriber 2: [ˈbiβo] for libro ‘book’).

Results

Translation equivalents

Before proceeding to an analysis of the phonological characteristics of the children’s target word forms and output forms, we examined the early lexicons of the bilingual children for the percentage of translation equivalents, that is, pairs of words which have the same meaning in both languages (e.g., German: Mond; Spanish: luna ‘moon’). The mean percentage of translation equivalents in the bilingual children’s first words was 30% (Simon: 25.6% [20/78]; Nils: 33.3% [32/96]; Jens: 29.7% [24/79]; Manuel: 29.7%

Kehoe 103

[27/91]), which is similar to the mean percentage documented by Pearson, Fernández, and Oller (1995) for Spanish–English bilingual children (aged between 0:8 and 2;6). Thus, the bilingual children in this study appear to be commensurate with other bilingual children in terms of their early semantic development.

Target word forms and output forms

Figures 2–7 display the percentages of target word forms exhibiting the six phonological characteristics for the four monolingual German and Spanish children and for the four bilingual children. The figures show both individual as well as group trends; each indi-vidual column represents a single child and each grouping of columns represents a dif-ferent group (monolingual German, monolingual Spanish, bilingual German, bilingual Spanish). Given the small number of participants and the potential number of contrasts to be studied which could result in experiment-wise error, we did not subject the data to statistical analyses. Instead we examined the data for ‘salient tendencies’ in terms of frequency trends. We consider a salient tendency to be one in which the highest value registered in the group with the lowest scores is still lower than the lowest value regis-tered in the other group (i.e., the group with the highest scores). We consider only those comparisons referred to in research questions 1 to 3 (see earlier), namely, comparisons between monolingual German versus monolingual Spanish, bilingual German versus bilingual Spanish, and monolingual versus bilingual children (i.e., monolingual German vs. bilingual German; monolingual Spanish vs. bilingual Spanish). The means and stand-ard deviations of the phonological characteristics of target word forms and output forms are given in Appendix 1.

In comparing monolingual German- and Spanish-speaking children, we observe sali-ent tendencies for the phonological characteristics: %3+syllables, %codas, and %velars. That is, all of the Spanish children selected longer words, fewer words with codas, and more words containing velar stops, than all of the German children. There were no sali-ent tendencies between German and Spanish monolingual children in the selection of words containing sibilants, laterals, or 2+PoA. Similar findings were observed in the comparison of the German and Spanish of the bilingual children. Bilingual children in their Spanish selected longer words, fewer words with codas, and more words containing velar stops than in their German. We did not observe any salient tendencies when com-paring monolingual and bilingual children. Overall, the findings indicated clear cross-linguistic differences between German- and Spanish-speaking children in the selection of words containing certain prosodic and segmental characteristics; however, these pat-terns were evident across both monolingual and bilingual children alike.

Figures 8–13 display the percentage of output forms exhibiting the six phonological characteristics for the four monolingual German and Spanish children and for the four bilingual children. In comparing monolingual German- and Spanish-speaking children, salient tendencies were observed for the variables %codas and laterals. Spanish-speaking monolingual children produced fewer words with codas and more words containing /l/ than the monolingual German-speaking children in their first words. There was overlap between groups (i.e., no salient tendencies) for the variables %3+syllables, %sibilants, %velars, and %2+POA. In the comparison of the German and Spanish of the bilingual children, no salient tendencies were obtained.

104 First Language 35(2)

010203040506070

Mon-Ger Mon-Sp Bi-Ger Bi-Sp

%Ta

rget

Wor

d Fo

rms

Figure 4. Percentages of target word forms containing sibilants for the monolingual German and Spanish and bilingual German–Spanish children. Each column indicates the results of a single child.

010203040506070

Mon-Ger Mon-Sp Bi-Ger Bi-Sp

%Ta

rget

Wor

d Fo

rms

Figure 2. Percentages of target word forms containing 3+syllables for the monolingual German and Spanish and bilingual German–Spanish children. Each column indicates the results of a single child.

010203040506070

Mon-Ger Mon-Sp Bi-Ger Bi-Sp

%T

arge

t W

ord

For

ms

Figure 3. Percentages of target word forms containing word-final codas for the monolingual German and Spanish and bilingual German–Spanish children. Each column indicates the results of a single child.

Kehoe 105

010203040506070

Mon-Ger Mon-Sp Bi-Ger Bi-Sp

%Ta

rget

Wor

d Fo

rms

Figure 5. Percentages of target word forms containing laterals for the monolingual German and Spanish and bilingual German–Spanish children. Each column indicates the results of a single child.

010203040506070

Mon-Ger Mon-Sp Bi-Ger Bi-Sp

%Ta

rget

Wor

d Fo

rms

Figure 6. Percentages of target word forms containing velars for the monolingual German and Spanish and bilingual German–Spanish children. Each column indicates the results of a single child.

010203040506070

Mon-Ger Mon-Sp Bi-Ger Bi-Sp

%T

arge

t W

ord

For

ms

Figure 7. Percentages of target word forms containing 2+PoA for the monolingual German and Spanish and bilingual German–Spanish children. Each column indicates the results of a single child.

106 First Language 35(2)

0102030405060

Mon-Ger Mon-Sp Bi-Ger Bi-Sp

% O

utpu

t For

ms

Figure 8. Percentages of output forms containing 3+syllables for the monolingual German and Spanish and bilingual German–Spanish children. Each column indicates the results of a single child.

0102030405060

Mon-Ger Mon-Sp Bi-Ger Bi-Sp

% O

utpu

t For

ms

Figure 9. Percentages of output forms containing codas for the monolingual German and Spanish and bilingual German–Spanish children. Each column indicates the results of a single child.

0102030405060

Mon-Ger Mon-Sp Bi-Ger Bi-Sp

% O

utpu

t For

ms

Figure 10. Percentages of output forms containing sibilants for the monolingual German and Spanish and bilingual German–Spanish children. Each column indicates the results of a single child.

Kehoe 107

0102030405060

Mon-Ger Mon-Sp Bi-Ger Bi-Sp

% O

utpu

t For

ms

Figure 11. Percentages of output forms containing laterals for the monolingual German and Spanish and bilingual German–Spanish children. Each column indicates the results of a single child.

0102030405060

Mon-Ger Mon-Sp Bi-Ger Bi-Sp

% O

utpu

t For

ms

Figure 12. Percentages of output forms containing velars for the monolingual German and Spanish and bilingual German–Spanish children. Each column indicates the results of a single child.

0102030405060

Mon-Ger Mon-Sp Bi-Ger Bi-Sp

% O

utpu

t For

ms

Figure 13. Percentages of output forms containing 2+PoA for the monolingual German and Spanish and bilingual German–Spanish children. Each column indicates the results of a single child.

108 First Language 35(2)

Some salient tendencies were observed when comparing the output forms of the monolingual and bilingual children. Bilingual children in Spanish produced more words with sibilants than Spanish monolingual children; bilingual children in German produced more words with laterals and variegated place than monolingual German children.

Additional tests

One possible reason as to why bilingual children produced more words with sibilants, /l/, and variegated place than monolingual children in one of their languages is meth-odological. The bilingual children were older than the monolingual children when they produced 50 words in both of their languages. To exclude the possibility that they had more advanced phonological abilities than the monolingual children because of their age, and that this contributed to the salient differences, we conducted additional analy-ses of the data. We reduced the age range of the bilingual data to make it comparable to the monolingual data. Instead of selecting 50 words types per language, we selected approximately 25–30 target word forms per language resulting in a cumulative total of 50–60 words across the two languages of the bilingual children. We then compared the reduced set of the bilingual children to a reduced set of the monolingual children (i.e., a set containing 25–30 words in each child), nevertheless, maintaining the same age range in the two groups. The analyses on the reduced set still showed that bilingual German children produced more laterals and variegated place than the monolingual German children but that not all bilingual Spanish children produced more sibilants than the monolingual Spanish children, although the tendency still remained (three of the bilingual Spanish children produced more sibilants than the four monolingual Spanish children). Thus, we do not believe that the bilingual children’s advanced age was the main reason for their better performance in some of the comparisons than the monolingual children’s. For the remainder of the article, we focus on the findings from the full analysis, that is, the one based on approximately 50 words per language for the bilingual children.

Whole-word templates

Monolingual children. Tables 2 and 3 display the whole word templates of the monolin-gual German- and Spanish-speaking children.

Consonant harmony processes were common in the first words of German-speaking children, occurring in approximately one-third of their productions (range: 26–39%). The most common harmony output patterns were ClabVClabV and Ct,dVCt,dV. A coda pat-tern was evident in the productions of all four German children: two children (Britta and Marion) produced a monosyllabic final-labial pattern (e.g., Arm [am], drehen [deːb], Mann [maːm], haben [ham], hopp [hɔpʰ]); one child (Bernd), a disyllabic final-/n/ pat-tern (e.g., baden [baːtɛn], duschen [dodæn], Ente [ʔatan], Teekanne [geðæn]), and one child (Thomas), a final-/t, d/ pattern, which was part of a more general Ct,dVCt,d(V) har-mony pattern (e.g., Dienstag [dit], Reifen [tatː]).

Another salient pattern amongst the German children was an /h/-initial pattern. Three of the four German children frequently selected /h/-initial words (e.g., haben, hallo,

Kehoe 109

Table 2. Templates in the first 50 target word forms of monolingual German children. The table displays frequency of templates and an example of a selected and an adapted form.

Child Template % Example selection Adaption

Birte 1. CHa 26% (14/53) ClabVClabV 15% (8/53) Baby [bebe] Birte [biba] 2. (C)VC 24% (13/53) (C)VClab 13% (7/53) ab [ʔap] haben [ham] 3. h-initial 19% (10/53) heiss [haɪç] Vogel [hoːgə]

Berndb 1. CH 26% (8/31) Ct,dVCt,dV 13% (4/31) Teddy [datiː] danke [dæda] 2. VCV 23% (7/31) alle [ʔajə] Schuhe [ʔɔwæ] 3. (C)VCVn 10% (3/31) duschen [dodæn] Purzel [bɔdan] 4. CVCVCV 10% (3/31) Angela [dɛlaʔa] bitte [pipipɛ]Marion 1. CH 39% (19/49) ClabVClabV 16% (8/49) papa [baba] Opa [bəpa] CnasVCnas(V) 16% (8/49) Mann [mam] Melone [noːɲo] 2. h-initial 14% (7/49) Haar [ha.a] Auto [hædʌ] 3. (C)VC 14% (7/49) (C)VClab 12% (6/49) ab [ʔap] oben [hom]

Thomas 1. CH 34% (16/47) ClabVClabVc 15% (7/47) Biber [biba] Ct,dVCt,d(V) 13% (6/47) tatütata [dada] Reifen [tatː] 2. (C)VkV 13% (6/47) Deckel [dakʰə] Rehbock [ʔeko] 3. h-initial 11% (5/47) heiss [has] Apfel [haɸa]

aCH = consonant harmony; examples are provided only when there is a consistent melodic pattern.b Bernd’s new target word forms were analyzed at 1;10.20. Bernd’s CH pattern was present at 22% (13/60) with a prevalent Ct,dVCt,dV pattern (18% [11/60]). His VCV pattern was present at 10% (6/60), his (C)VCVn pattern was present at 20% (12/60), and his CVCVCV pattern was present at 13% (8/60).

c Thomas’s ClabVClabV template was present only on the basis of selection (Biber, papa, piep piep, pitch patch, mama, wau wau, wuff wuff).

heiss, hopp hopp, hoppla, Haar, Hafer, Hasi, Hose) or adapted other words to /h/-initial output forms (Reisverschluss [hais], schlafen [haeɐ], Schlussel [hʏːʃɛ], schwer [heː], Licht [hɛːhi], Waschlappen [haja]). Bernd did not display the /h/-initial pattern but pro-duced a VCV pattern instead (e.g., Hasi [ʔatiː], hello [ʔaːɦoː], Schuhe [ʔɔwæ], Räder [ʔodaː]). The other templates in the German data were a multisyllabic template in Bernd’s productions (Angela [dɛlaʔa], ausziehen [ʔadeːʔɪ], aufstehen [ʔadeːɦən], Eisenbahn [ʔɛɪntɛnðæn]) and a medial /k/ pattern in Thomas’s productions (e.g., Deckel [dakʰə], Grekos [gikʰo], Onkel [oːkɐ], Raupe [aʊkʰoː]).

Consonant harmony was present in the Spanish monolingual children’s productions, although its frequency varied from as low as 18% to as high as 58%. All children dis-played a ClabVClabV pattern, but José, who had the most harmony patterns, also displayed coronal and dorsal harmony. Three of the four Spanish children evidenced a VCV pattern (e.g., agua [awa], alli [aji], aqui [ɐðiː], otro [ɔdʊ], diente [atæː], lapiz [ʔapɪ]) and one of

110 First Language 35(2)

these three children, María, also displayed a VCVCV pattern (e.g., Babero [ʔabeɾɔ], calcetin [adɛtiː], Laire [alalɛ], pesetita [ɛtito]). Two of the Spanish children produced an /h/-initial pattern. This pattern does not fulfill our definition of a template as it was pre-sent only through adaptation rather than selection (Spanish does not contain the phoneme /h/). We report its presence, nevertheless, to indicate that /h/-initial output patterns were not isolated to German. Two of the children displayed frequent use of velars either as a harmony pattern Ck,gVCk,gV or as a k-initial pattern. Two children also displayed melody patterns: CdorVCcorV (e.g., coche [kɔdə], conejo [kœno], cuento [gedoʊ], cuna [guna]) and ClabVCcorV (e.g., mano [mano], pala [pala], palo [palɔ], pato [bato]).

In sum, consonant harmony and the /h/-initial patterns were found in both German and Spanish and, thus, cannot be considered distinctive of either language. Similarly VCV and velar patterns were found across both languages, although they were more frequent in Spanish than in German. A multisyllabic pattern was evident in both lan-guages as well (Bernd and María).5 Templates that were language-specific were the coda patterns, which were found only in German (all four children), and the melody patterns which were found only in Spanish (two of the four children).

Table 3. Templates in the first 50 target word forms of monolingual Spanish children. This table displays frequency of templates and an example of a selected and an adapted form.

Child Template % Example selection Adaption

María 1. VCV 28% (17/61) VGVa 16% (10/61) allí [aji] mira [iːjɑ] 2. CHb 18% (11/61) ClabVClabV 10% (6/61) bebé [ßebe] chupete [pepe] 3. VCVCV 13% (8/61) a comer [ɐʎəʎæ] pesetita [ɛtito] 4. ClabVCcorV 10% (6/61) pala [pala] chupachup [baʧu]

Juan 1. CH 46% (23/50) ClabVClabV 18% (9/50) papá [babɐː] pelota [bæbɐ] 2. h-initialc 16% (8/50) sol [hol] 3. VCV 16% (8/50) otro [œdɔ] llama [amɐ]Miguel 1. CH 27% (12/44) ClabVClabV 11% (5/44) papá [papa] barco [bəbə] 2. k-initial 27% (12/44) coche [kɔdə] vaca [kaka] 3. VCV 14% (6/44) allí [ʔɐjɛ] lápiz [ʔapɪ] 4. CdorVCcorV 14% (6/44) cuna [guna] conejo [kœno]

José 1. CH 58% (23/40) ClabVClabV 15% (6/40) papá [bɐˈbɐ] pato [bɑpʰɐ] Ct,dVCt,dV 15% (6/40) dados [dadɐ] está [teːteː] Ck,gVCk,gV 12.5% (5/40) coco [kɔko] conejo [gœːgo] 2. h-initialc 22.5% (9/40) cinta [hita]

aVGV = vowel glide vowel.bCH = consonant harmony; examples are provided only when there is a consistent melodic pattern.ch-initial forms were present only as adapted forms since /h/ does not exist as a phoneme in Spanish.

Kehoe 111

Bilingual children. The whole-word templates identified for bilingual German- and Span-ish-speaking children are shown in Table 4. Interestingly, consonant harmony was not

Table 4. Templates in the first 50 target word forms of bilingual German–Spanish children. This table displays frequency of templates and an example of a selected and an adapted form.

Child Template % Example selection Adaption

Simon Gera 1. t,d-initial 31% (9/29) Tisch [tʏʃ] auf [taːfː] 2. p,b-initial 24% (7/29) Buch [buχ] weiter [baəta] 3. CHb 17% (5/29) Simon Sp 1. t,d-initial 35% (17/49) tres [tɛ] silla [tɪjːa] 2. CH 31% (15/49) tVta 20% (10/49) tarta [tata] botella [tʰetʰa] 3. Cp,bVCtV 12% (6/49) pato [patɔ] pelota [patʰə]

Nils Ger 1. VCV 16% (7/44) Affe [ʔafə] Hase [ɐӨɐ] 2. k-initial 16% (7/44) Kater [kɑːətʌ] Harke [kakə] 3. CH 11% (5/44) Nils Sp 1. ClabVCcorV 25% (13/51) pato [baːto] manzana [pada] 2. VCV 16% (8/51) oso [oso] caballo [aːlɔ] 3. (C)VlV 14% (7/51) cola [kola] pera [pela] 4. k-initialc 14% (7/51) queso [kɛsː] 5. CH 12% (6/51)

Jens Gerd 1. CVCc 22% (5/23) Bal [balː] 2. s-finalc 17% (4/23) Fisch [fɪs] 3. k-initial 13% (3/23) Kuh [kʊː] drei [kaɪ] 4. CH 13% (3/23) Jens Sp 1. CVkV 18% (10/56) vaca [baka] rojo [hoko] 2. 3 syllables 14% (8/56) pollitos [peˈsitos] perro [pəˈpɪlːʌ] 3. s-finalc 12.5% (7/56) patos [patos] 4. ClabVCcorV 12.5% (7/56) pato [pato] mariposa [posa] 5. CH 12.5% (7/56) 6. CVC 11% (6/56) sol [lʊl] mono [mʌn]Manuel Ger 1. ClabVCcor(V)c 24% (10/42) Bohne [bone] 2. CVCsib(V) 17% (7/42) was [vas] Kuchen [kʊse] 3. CH 12% (5/42) Manuel Sp 1. k-initial 17% (9/53) quiero [keðo] Till [kil] 2. VCV 17% (9/53) otra [ota] plato [ʔato] 3. (C)VCsib(V) 15% (8/53) coche [koʧe] pasta [paʧa] 4. ClabVCcor(V) 15% (8/53) mono [mano] vaca [bala]

a Simon’s new target word forms were analyzed at 1;11.06 in German. His t/d-initial pattern was present at 29% (6/21) and his b-initial pattern was present at 33% (7/21).

bCH = consonant harmony; examples are provided only when there is a consistent melodic pattern.cThese patterns were evident only in selection and not adaptation.d Jens’s new target word forms were analyzed in German at 1;11.28. He did not show frequent use of s-final or k-initial patterns in this session but he did evidence a new template: 29% (6/21) of output forms con-tained 3 syllables (e.g., Banane [baˈsane], Känguruh [tɬi.ti.ˈko:], Pinguin [po̞.kə.ˈtɪn:]).

112 First Language 35(2)

frequent in the bilingual data. It ranged from 11 to 13% in three of the four bilingual children, but was rarely associated with a consistent melodic pattern (e.g., ClabVClabV). It was slightly more frequent in one of the bilingual children’s productions, Simon, par-ticularly in his Spanish, where it took the form of a tVta template (e.g., tictac [tɪta], gal-letita [tita], quita [tita]). The /h/-initial pattern was not evident in the bilingual data.

VCV and velar patterns were found in the bilingual children. A VCV pattern was present in the productions of two children, Nils and Manuel. In Nils, it was present in both languages whereas in Manuel, only in Spanish. A velar pattern was found in three children. In Nils and Jens, it was present in both languages and in Manuel, it was present only in Spanish. Jens had a multisyllabic template (i.e., 3+syllables) in Spanish which was also present in German when the analysis was widened to include additional words (see footnote d in Table 4), although its presence in German may also be related to the older age range in which the data were taken.

Templates identified as language-specific on the basis of the monolingual analysis (e.g., the coda pattern in German and the melody pattern in Spanish) were also present in the bilingual children. A coda pattern was found in two of the bilingual children, Jens and Manuel. Jens’s coda pattern manifested in two ways: as a final /s/-pattern as well as a more general CVC form. Both patterns were present in the two languages. Manuel’s coda pattern was part of a more general sibilant pattern CVCsib(V), which was evident in both German and Spanish. A ClabVCcorV melody pattern was observed in all four children in their Spanish productions. In one child, Manuel, it was also present in German words through selection only.

The bilingual children also displayed templates that were not found in the monolin-gual data. Simon had /t, d/-initial and /p, b/-initial patterns, which were present across both German and Spanish. Similarly, Manuel had a medial (and final) sibilant pattern in his two languages (see above). In German, the sibilants were alveolar /s, z/ whereas in Spanish, they were alveolar-palatal fricatives and affricates /ʃ, ʧ/. In addition, Nils had a medial-/l/ pattern in Spanish (e.g., cola, bella, pala, hola) but not in German.

Summing up, bilingual children displayed similarities and differences to monolingual children in their whole word templates. In terms of similarities, bilingual children produced VCV, velar, and multisyllabic patterns in both German and Spanish as was seen in the monolingual data. In terms of differences, bilingual children did not display frequent use of consonant harmony and the /h/-initial pattern. Instead, they displayed templates that were not observed in the monolingual children such as sibilant and medial-/l/ templates. Bilingual children employed templates which were identified as being language-specific on the basis of the monolingual data, in both of their languages. For example, they used a coda pattern not only in German but also in Spanish.

Discussion

This study investigated the phonological composition of target word forms and word productions, as well as the presence of whole-word templates in the earliest lexicons of monolingual German- and Spanish-speaking children and in bilingual German–Spanish children. The general aim of the study was to examine lexical-phonological interactions in the early word period, with the specific aim of determining whether bilingual children

Kehoe 113

select different target word forms and produce different output forms and templates than monolingual children. In the following paragraphs, we summarize the findings in terms of our three research questions, review the findings on target word forms and output pat-terns (and templates) separately, and then consider the results in terms of the main theme, lexical-phonological interactions.

Summary of research findings

Summarizing the findings with respect to research questions 1 to 3, our analyses indi-cated that:

1. The phonological characteristics of target word forms, output forms, and whole-word templates did differ between German- and Spanish-speaking monolingual children. Spanish-speaking children selected more multisyllabic words and fewer words with codas than German-speaking children; they also produced fewer words with codas than the German-speaking children. Spanish-speaking children selected more words with velars, although they did not produce more words with velars. They produced more words with laterals, however. German-speaking children produced templates with codas, whereas Spanish-speaking children pro-duced ‘melody templates’, ones which were characterized by place variegation.

2. The phonological characteristics of target word forms differed between the two languages of the bilingual children. Similar to the findings in the monolingual data, bilingual children in Spanish selected more multisyllabic words, fewer words with codas, and more words with velars than they did in their German. The phonological characteristics of the output patterns did not differ greatly between the two languages. In addition, the templates in the two languages often resem-bled each other, although the phonetic manifestation was at times different between the two languages (e.g., alveolar sibilants in German vs. alveolar-palatal sibilants in Spanish as in Manuel’s sibilant template).

3. The phonological characteristics of output patterns and whole-word templates did differ between the monolingual and bilingual children but the phonological char-acteristics of target word forms did not differ. Bilingual children realized more laterals and variegated place in their German productions than the monolingual German children, and more sibilants in their Spanish productions than the mono-lingual Spanish children. Analogous effects were observed in whole-word tem-plates: the bilingual children’s templates were characterized by less frequent use of consonant harmony, and more frequent use of mature phonological patterns such as sibilants and laterals than the monolingual children’s templates. We now consider the findings on target word forms and output patterns in more detail.

Target word forms

Numerous studies have noted that there are salient prosodic differences between the Spanish lexicon versus the English and German lexicons (Demuth, 2011; Lleó & Demuth, 1999; Lleó et al., 2003; Roark & Demuth, 2000). Our results showed these prosodic

114 First Language 35(2)

differences to be present in the earliest target word forms of the monolingual and bilingual children. The monolingual and bilingual Spanish children selected more multisyllables and fewer words with word-final codas than the monolingual and bilingual German chil-dren, reflecting ambient language differences.

Our results indicated segmental differences between Spanish and German as well. The target word forms selected by the Spanish monolingual and bilingual children con-tained greater numbers of velars than the words selected by the German monolingual and bilingual children. The increased frequency of velars in Spanish compared to German is consistent with earlier observations by Lleó et al. (1996). They cite findings of Ortmann (1975), based on frequent word forms in German, and Quilis and Esgueva (1980), based on spoken speech in Spanish, which give a slight advantage to Spanish in terms of velar place (11.1% vs. 8.1%). The effect was quite prominent in the child data, with 16 words on average containing velars in the monolingual Spanish children’s word sets and 6 words containing velars in the monolingual German children’s word sets, suggesting that the Spanish child lexicon may contain an even higher percentage of velar consonants than that of adult speech.

The analysis of target word forms revealed salient cross-linguistic differences but few monolingual versus bilingual differences. That is, the bilingual children did not select more multisyllabic words in German than monolingual children did, due to the influence of Spanish. Thus, there does not seem to be any evidence for cross-linguistic interaction in the influence of phonology on the selection of words, a finding which will be dis-cussed in more detail below.

Output patterns and templates

Paralleling the analysis of target word forms, some salient cross-linguistic differences were evident in the word productions of the monolingual children. Spanish-speaking children produced fewer words with word-final codas, and more words with laterals than German-speaking children. In the analysis of whole-word templates, differences were observed in the production of codas and in place variegation: monolingual German chil-dren displayed more templates with codas and monolingual Spanish children displayed more melody templates. These results are consistent with numerous studies which report salient cross-linguistic differences between Spanish versus English/German children in their production patterns (Cataño et al., 2009; Lleó, 2008; Lleó & Demuth, 1999; Lleó et al., 1996, 2003). Specifically, Lleó et al. (1996), using a similar database to the current one, found that German-speaking children produced significantly more codas than Spanish-speaking children at the 25-word point; and Cataño et al. (2009) found that /l/ was present in the early phonetic inventories of Spanish-speaking children but not in those of English-speaking children, a finding that is relevant to this study, since German may pattern similarly to English in the development of laterals. Indeed, our results indi-cated that, on average, /l/ was present in one word of the monolingual German-speaking children’s productions versus five words of the Spanish-speaking children’s productions. Cataño et al. (2009) explored several different reasons for the early appearance of /l/ in Spanish compared to English, including articulatory and allophonic differences, fre-quency of occurrence, and functional load. The authors hone in on the fact that /l/ has a

Kehoe 115

greater functional load in Spanish than it has in English because it occurs in many arti-cles and clitic pronouns (e.g., el, la, los, las, le, lo, les). This may also explain its higher frequency in the early production forms of Spanish compared to German. Important to note is that /l/ was not more frequent in the target word forms selected by Spanish- versus German-speaking children.

Interestingly, Spanish-speaking monolingual children selected more multisyllables and words with velars than German-speaking monolingual children; however they did not produce more of these structures in their first words than German-speaking chil-dren. In the case of multisyllabic words, there was a tendency for Spanish-speaking monolinguals to produce more of them than the German-speaking monolinguals, but there was overlap between groups with one German-speaking child (Bernd) also pro-ducing a number of multisyllabic words. Thus, the small numbers of multisyllabic words produced by the children combined with individual differences obscured salient tendencies. In the case of velars, differences between groups were hardly evident. Velars were subject to individual differences within languages, with some children producing several velars in their first words and others very few. Marion (represented by column 3 in the monolingual German group in Figures 6 and 12) did not produce any words with velars nor did she select many words with velars in contrast to the other German monolingual children who selected some. Thus, we find evidence for selection and avoidance of velars but on an idiosyncratic rather than on a language-specific basis. If we had sampled words beyond the 25-word point, significant differences in the produc-tion of velars may have been observed between the Spanish and German monolingual children (Lleó et al., 1996).

Turning to the bilingual comparisons, we observed few cross-linguistic differences in the output patterns of the two languages of bilingual children, and we observed salient bilingual–monolingual differences in terms of certain segmental features (e.g., /l/, and place variegation). Bilingual children displayed less use of consonant harmony and /h/ than the monolingual children; these are patterns that might be expected in early phono-logical systems. They displayed greater use of templates containing sibilants, laterals, and variegated place than the monolingual children; these are patterns that might be expected in more mature phonological systems. Combining the various analyses on pro-duction forms, two main results emerged: (1) there were fewer cross-linguistic differ-ences between the two languages of the bilinguals than there were between the two respective monolingual populations; and (2) there was a tendency for bilingual children’s productions to be phonetically/phonologically more sophisticated than those of the monolingual children’s in certain segmental variables. In the following section, we explore the bilingual children’s phonological results in more detail.

Bilingual children’s phonology

Numerous studies provide evidence for early differentiation as well as interaction between the phonological systems of bilingual children (Fabiano-Smith & Goldstein, 2010; Kehoe, Lleó, & Rakow, 2004; Keshavarz & Ingram, 2002; Montanari, 2011). In the current study, the phonological differences that were present in the monolingual children’s data were not present to the same degree in the bilingual data, suggesting potential instances of

116 First Language 35(2)

cross-linguistic interaction. For example, the monolingual children produced more /l/s in Spanish than in German; yet, the Spanish productions of the bilingual children did not contain greater numbers of /l/s compared to their German productions.6 Similarly, the monolingual children produced more codas in German than in Spanish; yet, the Spanish productions of the bilingual children also contained quite a number of codas making the differences between groups not as extreme as in the monolingual data. The production of these phonological features may have come about through the influence of the other lan-guage. The production of /l/s in Spanish may have facilitated the production of them in German. A similar effect may have been present for codas, although the effect was subtle since salient tendencies were not evident when comparing the bilingual and monolingual children in Spanish (as was seen when comparing bilingual and monolingual children in the production of /l/ in German). Studies by our research group, however, have found greater production of codas by bilingual children in Spanish in comparison to monolin-gual children in Spanish at later stages of development (Lleó et al., 2003), suggesting that cross-linguistic interaction effects for codas become stronger after the first word period.

Nevertheless, some of the findings are less easy to account for in terms of cross-lin-guistic interaction. The bilingual children produced greater numbers of words containing sibilants and place variegation than the monolingual children, yet place variegation and sibilants were not more frequent in the target word forms of one language versus the other.7 Here, we posit that the results reflect a ‘more general effect’ of bilingualism on phonological development. A number of studies report superior phonetic and phonologi-cal abilities in bilingual as compared to monolingual children (Grech & Dodd, 2008; Johnson & Lancaster, 1998; Montanari, 2011). These studies have measured general aspects of phonology such as size of the phonetic inventory and percent consonants cor-rect and have found them to be more advanced in bilingual compared to monolingual children. Johnson and Lancaster (1998) suggest that bilingualism may lead to heightened attention to the phonemic contrasts of the two languages, which in turn results in the earlier production of sounds unique to each language, some of which are typically late acquired or more marked. In a similar vein, Grech and Dodd (2008) hypothesize that the act of distinguishing between two languages enhances bilingual children’s awareness of language-specific phonological constraints and, ultimately, leads to their increased pho-nological knowledge. If we consider the current findings on ‘place variegation’, we observe that the bilingual children displayed less use of consonant harmony and greater use of place variegation than the monolingual children. Given that certain authors pro-pose that consonant harmony reflects an early stage in phonological development when children’s phonological representations are unspecified (Fikkert & Levelt, 2008), our findings could suggest that the bilingual children’s phonological representations are more complete/specified than the monolingual children’s at a similar stage of word development. This increased phonological knowledge would come from exposure to a greater number of phonological contrasts present in the early target word forms selected across both of the bilingual children’s languages.

Since the study is based on small group sizes (n = 4), caution should be exerted in the interpretation of the results. The results could be due to individual differences or random factors. We may have inadvertently selected bilingual children who were more advanced phonologically than the monolingual children. Although we cannot fully exclude this possibility, it seems unlikely, given that studies on the same group of bilingual children

Kehoe 117

have not documented superior phonetic/phonological abilities in other areas of phono-logical development. For example, the bilingual children have not been found to be more advanced than the monolingual children in their acquisition of Voice Onset Time (VOT) (Kehoe et al., 2004), the vowel length contrast (Kehoe, 2002), and rhythm (Kehoe, Lleó, & Rakow, 2011). Admittedly, these studies were conducted when the children were older, that is, aged between 2 and 3 years. However, if the findings were due to individual differences amongst children, we might have expected superior results with this group of children across other domains even at a later stage of development. Socio-economic status and parental education levels may also influence children’s lexical and phonologi-cal development (Campbell et al., 2003; Hoff, 2003); nevertheless, there was no evi-dence that these factors played a role in the monolingual–bilingual differences since the two groups were homogeneous in terms of these factors. Finally, even though we con-ducted additional analyses in which we controlled for the bilingual children’s advanced age at the time of first word production, it still must be conceded that the bilingual chil-dren were on average a little older and their cumulative lexicons were slightly larger than the monolingual children’s. Whether the small differences in age and cumulative vocab-ulary of the bilinguals in comparison to the monolinguals account for their advanced phonology remains a question for future research.

A closer look at templates in bilingual children

Apart from some individual case studies by Vihman (2002, 2009), findings on templates in bilingual children are scarce. In addition, there are few studies on templates in German- and Spanish-speaking children. Thus, an important aspect of the current study was the analysis of templates in these populations. This study attempted to formalize the measure of templates by requiring that templates be reasonably frequent, be present in both selec-tion and adaptation processes, and be defined as narrowly as possible. However, even given such criteria, the identification of templates was not always clear cut. The use of a more conservative frequency criterion (20% vs. 10%) might have led to the identification of more salient templates, but would have meant that the majority of children would not have been credited as having templates. Indeed, the latter possibility may represent the reality, since findings on templates are based predominantly on individual case studies which may amplify the presence of templates in the population at large. The ‘operation-alization’ of templates and the identification of them in a large-scale study should be seen as priority for this approach.

Leaving aside methodological difficulties associated with the identification of tem-plates, our findings agree with Vihman’s (2002, 2009) in showing that bilingual children tend to have similar templates in both languages. VCV, coda, and velar templates often appeared in both languages of the bilingual children. However, the templates did not always look exactly the same across languages. A case in point is Simon’s tVta pattern (see Table 5). This pattern appeared to be part of a general /t, d/-initial pattern. Simon selected /t, d/-initial words in German and Spanish (e.g., da, Teddy, danke, Tisch, tictac, tarta, tres); he inserted /t, d/ as a default onset in German (Auto [dado], auf [taːfː]) and he substituted /t, d/ for other sounds in both German and Spanish (Zunge [dɔkɛ], Löwe [dœːvɛ], calle [tajɛ], silla [tɪjːa]); however, only in Spanish did the template appear as a

118 First Language 35(2)

tVta form. Simon was most likely inspired by the diminutive form in Spanish which is a salient and highly frequent characteristic of input to young children (e.g., casita ‘little home’, abuelita ‘little grandmother’, galletita ‘little biscuit’). He produced the tVta form for target dimunitives (e.g., galletita and barriguita) as well as non-diminutives (e.g., galleta, quita). The German word danke /ˈdaŋkə/ did not evoke the tVta pattern, either because it did not resemble the phonetic pattern common to the Spanish target words, which consisted predominantly of dorsal …coronal, or because it belonged to a target language in which the diminutive form did not exist. In either case, the findings suggest that the pattern induction made by the bilingual children, when selecting and adapting target forms to templates, pays attention to fine language-specific details.

In some cases, similar templates were not observed in the two languages of the bilin-gual children such as the example of Nils’ melody template <ClabVCcorV>, which appeared only in Spanish (see Table 6). A closer examination of the data reveals that German offered very few input forms adhering to the labial-coronal melody pattern.

Table 5. Templates present across both languages but with different phonetic manifestations: </t, d/-initial> patterns in Simon’s productions.

Age German Spanish

Selection Adaptation Selection Adaptation

Simon </t, d/-initial>, <tVta> in Spanish

1;04 da [dɐ] tictac [tɪta] calle [tajɛ] Teddy [dɛdɪ] die [di]

1;05 Auto [dado]

1;06 tarta [tata] tres [tɛ] 1;07 danke [takɛ] hier [diɐ] tuya [tia] galletita [tita] quita [tita] galleta [teta] gracias [təɐta]1;08 Auf [taːfː] silla [tɪjːa] barriguita [tʰɪtʰa]1;09 barriga [tʰɪtʰa] botella [tʰetʰa] anillo [tʰɪːʝo] locomotora [tɔːla] osito [dɪto]1;10 Tisch [tʏʃ] Stein [dae]

1;11 das [das] Zunge [dɔkɛ] Tiger [tɪgɣɐː] Zug [tuk] Löwe [dœːvɛ]

Kehoe 119

Apart from the isolated occurrences listed in Table 6 (Bälle, Wasser, meiner), there was only one other example in Nils repertoire, bitte, which he produced as [tɛː]. In contrast, Spanish offered many more <ClabVCcorV> input forms and the possibility to create them via truncation (manzana -> mana). Thus, the findings suggest that templates are con-strained by the phonological characteristics of the ambient language and large differ-ences between languages reduce the chances of seeing common templates in both languages (see Keren-Portnoy et al., 2009, who observe melody templates in Italian-speaking children based on frequent input patterns). The fact that bilingual children do not always produce similar templates in their two languages provides support that they phonologically differentiate their languages.

Lexical-phonological interactions

Finally, we return to the larger question of the study, namely whether the lexical selection and output patterns (plus whole-word templates) of bilingual children differ from those of monolingual children. Our results suggest that output (and templatic) patterns differ but not lexical selection patterns. In terms of the possibilities displayed in Figure 1, our findings support number 3. We can most clearly eliminate Possibility 2 because the pres-ence of salient monolingual–bilingual differences suggests that the two phonological and lexical systems of bilinguals are not entirely independent. At first glance, the lack of language-specific differences between the output forms of bilingual children suggests that a single phonological strategy could be the source of early phonological-lexical interactions, thus implicating Possibility 1. However, not all phonological characteristics were realized to the same degree in the two languages of the bilingual children, suggest-ing differentiation between the two phonological systems as well. Our findings are most consistent with independence and interaction between phonological systems, but not in all directions, thus, implicating a modified version of Possibility 3.

Table 6. Template present in one language only: <ClabVCcorV> template in Nils’ Spanish productions (only isolated examples of this pattern are present in German).

Age German Spanish

Selection Adaptation Selection Adaptation

<ClabVCcorV>

1;03 pato [baːto] zapato [paːtɔ]1;04 pala [bɔlːaː]

1;05 mano [manːœ] Mono [ˈmʌˈno] 1;06 Bälle [bɛjɐ] pera [pela] Wasser [basa] elefante [paːteː] mermelada [mata] manzana [pada]1;07 meiner [maɪnə] paté [pata] mesa [pəsɐ] 1;08 Bella [bəlɐː] cepillo [piʎo]

120 First Language 35(2)

Let us consider some of the main findings of the study in terms of lexical- phonological interactions. The lack of salient differences in the productions of laterals and codas between the two languages of the bilinguals is suggestive of interaction between the two phonological systems, and of ambient language effects of one lexicon on the phonology of the other. Codas are frequent in German words. Laterals are not more frequent in the target word forms of Spanish (at least in the ones selected by the children) but the fact that they occur in high frequency function words suggests that they are a salient aspect of Spanish input. We assume that the high frequency of codas and laterals in one lan-guage influences its production both in that language and in the other language. The speech-motor component of phonology may facilitate this process. Sounds and sound sequences which are practiced in one language may be more easily produced in another.

What about interaction in the opposite direction? Did the study provide evidence of phonological influences on the bilingual children’s lexical selection patterns? The answer is no. The bilingual children selected similar types of words to the monolingual children in terms of their phonological composition. For example, they selected several multisyl-labic words in Spanish, but they did not select more multisyllabic words in German than the German monolingual children due to the influence of Spanish.

Why cross-linguistic interaction was observed in output forms but not in word selec-tion is unclear but it could be hypothesized that it relates to modality differences between phonological and lexical domains. Interaction between phonological systems may be more direct than between lexical systems because it is mediated by the child’s own pro-ductive and articulatory capacities. Indeed studies examining lexical-grammatical asso-ciations in bilingual children have found stronger evidence for within- than cross-language correlations (Conboy & Thal, 2006; Marchman, Martínez-Sussmann, & Dale, 2004), sug-gesting that the lexical (and grammatical) domains of a given language may be more autonomous in bilingual children than the phonological domains. More recently, Cooperson, Bedore, and Peña (2013) have reported significant between-language correla-tions when examining the phonological and language skills of bilingual children, suggest-ing that phonology may operate differently from other domains of language. The current findings also provide support for a between-language component to the phonology– language (lexicon) relationship in bilingual children, but one that goes in one direction only.

Finally, as mentioned earlier, we do not know the extent to which phonology influ-ences early word selection even in the monolingual child. The impact of phonology may still be relatively weak when compared to pragmatic, semantic, or word frequency effects on word selection. This weakness may be compounded in the bilingual situation when other factors come into play such as language dominance, language context, and the language background of the child’s interlocutors.

Conclusion

An analysis of the early target word forms and word productions of monolingual and bilingual German- and Spanish-speaking children indicates that bilingual children’s word productions may be characterized by more marked segmental structures, such as increased use of /l/ and variegated place. We propose that this effect arises as a result of

Kehoe 121

cross-linguistic interaction or is due to a more general effect of bilingualism which pro-motes increased phonetic differentiation. These findings are important since they show cross-linguistic influence on bilingual children’s early word productions even during the first word period. Bilingual children’s selection of target word forms did not differ from monolingual children’s, which suggests that the lexical domain is less prone to cross-linguistic interaction than the phonological domain.

Funding and acknowledgements

The monolingual data were collected as part of the project PAIDUS, financed by the DFG (German Science Foundation) and the DAAD (Acciones Integradas). The bilingual data were collected at the Research Centre on Multilingualism in Hamburg, also supported by the DFG. Both projects were directed by Professor Conxita Lleó. I would like to thank the children and parents who took part in the study as well as David Warne for his assistance in data analysis.

Notes

1. Figure 1 shows interaction between the lexical systems in Possibility 3; however, this may take place even with or without interaction between the phonological systems. Hence, lexi-cal interaction may be present in Possibilities 1 and 2 as well. The current study focuses on interaction between phonological systems and between the lexical-phonological interface and not between lexical systems per se.

2. Whole-word templates are output forms as well, but we refer to them separately, because they were subject to different analyses, namely, qualitative rather than quantitative.

3. It is possible that the 50-word period for the bilingual children is best measured in terms of total vocabulary across the two languages. Thus, it may have been preferable to count 25 target word forms in each language. For statistical purposes, we aimed for comparable numbers across monolingual and bilingual groups. Therefore we still chose 50 target word forms per language where possible. An analysis which concentrated on a cumulative total of 50 words across the two languages of the bilinguals produced essentially similar results (see section Additional tests).

4. We also examined the percentage of monosyllables in the German and Spanish data but since it provided complementary results to the percentage of 3+syllables, we present the findings of one of these variables.

5. We acknowledge that even though similar templates were observed in Spanish and German (e.g., Bernd and María both produced long words), the finer aspects of them could be quite different. For example, Bernd primarily selected onset-less long words (e.g., aufziehen [ʔadeːʔi]; aufstehen [ʔadeːɦən]) which was not the case with María (e.g., rastrillo [atiːja]; pesetita [ɛtito]).

6. One of the bilingual children, Simon, did not produce many /l/s in his early Spanish words (see Figure 11, column 1 under Bi-Ger and Bi-Sp), although he produced several in his German words. In the session following the 25-word point, Simon produced many words with /l/ (e.g., campana [ˈpala], pala [ˈpala], luna [ˈlʊnba], plato [ˈpalo], tigre [ˈtigle], verde [ˈbeleː]). Thus, the absence of them in Spanish may have been due to sampling issues.

7. It is possible that the target word forms selected by Spanish children contain more place variegation than those selected by German children by virtue of the fact that they are multisyl-labic and contain a greater number of velars. This may explain why more melody templates were observed in the productions of Spanish- compared to German-speaking monolingual children.

122 First Language 35(2)

References

Bernhardt, B., & Stoel-Gammon, C. (1996). Underspecification and markedness in normal and disordered phonological development. In C. Johnson & J. Gilbert (Eds.), Children’s language (Vol. 9, pp. 33–53). Mahwah, NJ: Lawrence Erlbaum.

Campbell, T., Dollaghan, C., Rockette, H., Paradise, J., Feldman, H., Shriberg, L., Sabo, D., & Kurs-Lasky, M. (2003). Risk factors for speech delay of unknown origin in 3-year-old chil-dren. Child Development, 74, 346–357.

Cataño, L., Barlow, J., & Moyna, M. (2009). A retrospective study of phonetic inventory complex-ity in acquisition of Spanish: Implications for phonological universals. Clinical Linguistics and Phonetics, 23, 446–472.

Celce-Murcia, M. (1978). The simultaneous acquisition of English and French in a two-year-old child. In E. Hatch (Ed.), Second language acquisition: A book of readings (pp. 38–53). Rowley, MA: Newbury House Publishers.

Conboy, B., & Thal, D. (2006). Ties between the lexicon and grammar: Cross-sectional and longi-tudinal studies of bilingual toddlers. Child Development, 77, 712–735.

Cooperson, S., Bedore, L., & Peña, E. (2013). The relationship of phonological skills to language skills in Spanish-English-speaking bilingual children. Clinical Linguistics and Phonetics, 27, 371–389.

Delattre, P. (1966). A comparison of syllable length conditioning among languages. International Review of Applied Linguistics, 4, 183–198.

Demuth, K. (2011). Interaction between lexical and phonological development; cross-linguistic and contextual considerations – A commentary on Stoel-Gammon’s ‘Relationships between lexical and phonological development in young children’. Journal of Child Language, 38, 69–74.

Fabiano-Smith, L., & Goldstein, B. (2010). Phonological acquisition in bilingual Spanish-English speaking children. Journal of Speech, Language and Hearing Research, 53, 160–178.

Ferguson, C., & Farwell, C. (1975). Words and sounds in early language acquisition. Language, 51, 419–439.

Fikkert, P., & Levelt, C. (2008). How does place fall into place? The lexicon and emergent con-straints in the developing phonological grammar. In P. Avery, B. E. Dresher & K. Rice (Eds.), Contrast in phonology: Perception and acquisition (pp. 219–256). Berlin, Germany: Mouton.

Grech, H., & Dodd, B. (2008). Phonological acquisition in Malta: A bilingual language learning context. International Journal of Bilingualism, 12, 155–171.

Grijzenhout, J., & Joppen, S. (1999). First steps in the acquisition of German phonology: A case study (Rutgers Optimality Archives. ROA-304-0399). Retrieved from http://roa.rutgers.edu/article/view/314

Hoff, E. (2003). The specificity of environmental influence: Socioeconomic status affects early vocabulary development via maternal speech. Child Development, 74, 1368–1378.

Ingram, D. (1981). The emerging phonological system of an Italian-English bilingual child. Journal of Italian Linguistics, 2, 95–113.

Ingram, D. (2002). The measurement of whole-word productions. Journal of Child Language, 29, 713–733.

Jakobson, R. (1968). Child language, aphasia, and phonological universals. The Hague, The Netherlands: Mouton. (Originally published as Kindersprache, aphasie und allgemeine laut-gesetze, Uppsala, Sweden, 1941).

Johnson, C., & Lancaster, P. (1998). The development of more than one phonology: A case study of a Norwegian-English bilingual child. International Journal of Bilingualism, 2, 265–300.

Kehoe, M. (2002). Developing vowel systems as a window to bilingual phonology. International Journal of Bilingualism, 6, 315–334.

Kehoe 123

Kehoe, M., Lleó, C., & Rakow, M. (2004). Voice onset time in bilingual German-Spanish chil-dren. Bilingualism: Language and Cognition, 7, 71–88.

Kehoe, M., Lleó, C., & Rakow, M. (2011). Speech rhythm in the pronunciation of German and Spanish monolingual and German-Spanish bilingual 3-year-olds. Linguistische Berichte, 227, 323–351.

Keren-Portnoy, T., Majorano, M., & Vihman, M. (2009). From phonetics to phonology: The emer-gence of first words in Italian. Journal of Child Language, 36, 235–265.

Keshavarz, M., & Ingram, D. (2002). The early phonological development of a Farsi- English bilingual child. International Journal of Bilingualism, 6, 255–269.

Lleó, C. (2006). The acquisition of prosodic word structures in Spanish by monolingual and Spanish-German bilingual children. Language and Speech, 49, 205–229.

Lleó, C. (2008). Research on first language acquisition of Spanish phonology. Studies in Hispanic and Lusophone Linguistics, 1, 349–384.

Lleó, C., & Demuth, K. (1999). Prosodic constraints on the emergence of grammatical morphemes: Cross-linguistic evidence from Germanic and Romance languages. In A. Greenhill, H. Littlefield, & C. Tano (Eds.), Proceedings of the 23rd Annual Boston University Conference on Language Development (pp. 407–418). Somerville, MA: Cascadilla Press.

Lleó, C., & Kehoe, M. (2002). On the interaction of phonological systems in child bilingual acqui-sition. International Journal of Bilingualism, 6, 233–237.

Lleó, C., Kuchenbrandt, I., Kehoe, M., & Trujillo, C. (2003). Syllable final consonants in Spanish and German monolingual and bilingual acquisition. In N. Müller (Ed.), (Non)vulnerable domains in bilingualism (pp. 191–220). Amsterdam, The Netherlands: John Benjamins.

Lleó, C., Prinz, M., El Mogharbel, C., & Maldonado, A. (1996). Early phonological acquisi-tion of German and Spanish: A reinterpretation of the continuity issue within the principles and parameters model. In C. E. Johnson & J. Gilbert (Eds.), Children’s language (Vol. 9, pp. 11–31). Mahwah, NJ: Lawrence Erlbaum.

Macken, M. (1979). Developmental reorganization of phonology. Lingua, 49, 11–49.Majorano, M., & D’Odorico, L. (2011). The transition into ambient language: A longitudinal study

of babbling and first word production of Italian children. First Language, 31, 47–66.Marchman, V., Martínez-Sussmann, & Dale, P. (2004). The language-specific nature of gram-

matical development: Evidence from bilingual language learners. Developmental Science, 7, 212–224.

Menn, L. (1976). Pattern, control, and contrast in beginning speech (Unpublished PhD thesis). University of Illinois, Chicago, IL.

Menn, L. (1983). Development of articulatory, phonetic, and phonological capabilities. In B. Butterworth (Ed.), Language production (Vol. 2, pp. 3–49). London, England: Academic Press.

Montanari, S. (2011). Phonological differentiation before age two in a Tagalog-Spanish-English trilingual child. International Journal of Multilingualism, 8(1), 5–21.

Ortmann, W. (1975). Beispielwörter für deutsche Ausspracheübungen. 7591 hochfrequente Wortformen der Kaeding-Zählung, rechnersortiert nach Einzellauten, Lautverbindungen, Silbenzahl and Akzentposition [Examples of words for German pronunciation drills. 7951 highly frequent word forms of the Kaeding count, computerized according to single sounds, sound combinations, numbers of syllables and stress position]. München, Germany: Goethe-Institut.

Paradis, J. (2000). Beyond ‘One System or Two?’ Degrees of separation between the languages of French-English bilingual children. In S. Döpke (Ed.), Cross-linguistic structures in simulta-neous bilingualism (pp. 175–200). Amsterdam, The Netherlands: John Benjamins.

124 First Language 35(2)

Pearson, B. Z., Fernández, S., & Oller, D. K. (1995). Cross-language synonyms in the lexicons of bilingual infants: One language or two? Journal of Child Language, 22, 345–368.

Quilis, A., & Esgueva, M. (1980). Frecuencia de fonemas en el español hablado [Phoneme fre-quencies in spoken Spanish]. Lingüística Española Actual, II, 1–25.

Roark, B., & Demuth, K. (2000). Prosodic constraints and the learner’s environment: A cor-pus study. In S. Howell, S. Fish, & T. Keith-Lucas (Eds.), Proceedings of the 24th Annual Boston University Conference on Language Development (pp. 597–608). Somerville, MA: Cascadilla Press.

Schwartz, R., & Leonard, L. (1982). Do children pick and choose? An examination of phono-logical selection and avoidance in early lexical acquisition. Journal of Child Language, 9, 319–336.

Stoel-Gammon, C. (2011). Relationships between lexical and phonological development in young children. Journal of Child Language, 38, 1–34.

Stoel-Gammon, C., & Cooper, J. (1984). Patterns of early lexical and phonological development. Journal of Child Language, 11, 247–271.

Stoel-Gammon, C., & Dunn, K. (1985). Normal and disordered phonology in children. Austin, TX: Pro-Ed.

Velleman, S., & Vihman, M. (2002). Whole-word phonology and templates. Language, Speech, and Hearing Services in Schools, 33, 9–23.

Vihman, M. (1978). Consonant harmony: Its scope and function in child language. In J. Greenberg (Ed.), Universals of human language (pp. 281–334). Stanford, CA: Stanford University Press.

Vihman, M. (2002). Getting started without a system: From phonetics to phonology in bilingual development. International Journal of Bilingualism, 6, 239–254.

Vihman, M. (2009, November 16). Templates in phonological development: The case of bilin-guals. Paper presented at Bangor University, Bangor, UK.

Vihman, M. (2010, January 28–30). Templates in adult and child language. Paper presented at the Workshop on Templates, OCP 7, Nice, France.

Vihman, M., & Croft, W. (2007). Phonological development: Toward a ‘radical’ templatic phonol-ogy. Linguistics, 45, 683–725.

Vihman, M., & Keren-Portnoy, T. (2013). The emergence of phonology: Whole-word approaches and cross-linguistic evidence. Cambridge, UK: Cambridge University Press.

Vihman, M., Macken, M., Miller, R., Simmons, H., & Miller, J. (1985). From babbling to speech: A re-assessment of the continuity issue. Language, 61, 397–445.

Vihman, M., & Velleman, S. (2000). The construction of a first phonology. Phonetica, 57, 255–266.Waterson, N. (1971). Child phonology: A prosodic view. Journal of Linguistics, 7, 179–211.Yavas, M. (1995). Phonological selectivity in the first fifty words of a bilingual child. Language

and Speech, 38, 189–202.

Kehoe 125

%3+syll %codas %sib %lat %vel %2+PoA

M SD M SD M SD M SD M SD M SD

Target word formsMonGer 10.5 6.9 43.5 8.3 21.5 5.0 16.7 3.0 13.3 6.6 42.6 5.5MonSp 26.4 2.1 20.5 5.8 18.0 5.6 20.1 3.6 33.1 2.9 54.5 7.0BiGer 3.2 3.0 47.0 6.4 26.8 4.2 14.1 5.3 13.3 5.0 48.4 5.5BiSp 28.4 7.2 17.6 7.4 22.4 4.2 17.6 3.6 30.6 1.1 61.0 6.3

Output patternsMonGer 3.0 4.6 20.5 8.0 8.5 8.7 2.4 1.8 8.8 8.4 6.7 5.3MonSp 12.5 6.2 3.1 4.7 1.5 1.7 10.5 1.8 15.4 9.6 19.9 11.0BiGer 2.6 3.4 31.8 12.4 21.7 12.4 12.4 6.1 11.9 4.9 34.5 4.3BiSp 13.3 9.2 10.1 9.7 18.2 13.7 14.6 7.3 19.8 9.3 46.1 15.7

Appendix 1. Means and standard deviations of phonological characteristics of target word forms and output forms.