Language-specific influence on phoneme development: French and drehu data

Julia Monnin, Hélène LœvenbruckInterspeech 2010

TL;DR

This study investigates how the frequency of sounds in a child's native language influences their early phoneme development. By comparing French and Drehu speaking children, researchers found that sound frequency strongly predicts early production accuracy, especially in younger children, suggesting a significant role for linguistic input in shaping phonological acquisition.

What Problem Does This Paper Address?

Understanding how children acquire language is a fundamental question in linguistics and cognitive science. While it's known that children follow universal patterns in phoneme development (e.g., some sounds are inherently harder to produce), the extent to which their specific linguistic environment shapes this process is less clear. This paper addresses the question of how language-specific distributional patterns of phonemes and sound sequences influence a child's mastery of these sounds, aiming to disentangle universal phonetic constraints from language-specific learning.

Methods

The researchers employed a cross-linguistic approach, comparing phoneme production in French- and Drehu-acquiring children. They used two main methods:

  • Word and Non-word Repetition Tasks: Children were asked to repeat words and made-up "non-words." This allowed for the analysis of both familiar (word) and novel (non-word) sound sequences.
  • Correlation Analysis: They correlated the children's accuracy in producing specific word-initial consonant-vowel (CV) sequences with the frequency of those sequences in adult speech (French and Drehu corpora).
  • Statistical Analysis (ANOVA): Analysis of Variance was used to compare production scores between language groups and across different age groups, controlling for factors like age and language.

The novelty lies in the direct comparison of production accuracy against input frequency in two distinct linguistic environments, particularly including Drehu, a less commonly studied language in this context.

Datasets

The study utilized two primary types of data:

  • Adult Speech Corpora: Frequency data for phonemes and CV sequences were derived from recordings of adult speech directed at young children in both French and Drehu. This represents the "input" children are exposed to.
  • Child Production Data: Recordings of children aged 2 to 5 years repeating words and non-words in French and Drehu. This data was collected from:
  • Monolingual French-acquiring children.
  • Drehu-acquiring children in a bilingual French-Drehu environment (referred to as "bilingual children").

The study also drew upon existing research and data from other languages (Cantonese, English, Greek, Japanese) as part of the broader Paidologos Project.

Key Insights and Ideas

  • Frequency Matters, Especially Early On: The study strongly supports the idea that the frequency of a sound or sound sequence in a child's ambient language significantly influences their production accuracy. This effect is most pronounced in younger children (2-3 years old).
  • Language-Specific Patterns Emerge: While universal tendencies exist, the specific sounds that are acquired earlier or later can be influenced by their frequency in the target language. For example, sounds rare in French but common in Drehu were better produced by bilingual children.
  • Input Frequency vs. "Difficulty": The research challenges the notion of universally "difficult" phonemes. For instance, the consonant /k/, often cited as difficult, was produced with high accuracy by both French and Drehu learners, likely due to its high frequency in both languages.
  • Bilingualism and Input Influence: Bilingual children showed advantages in producing sounds that are frequent in Drehu but rare in French, suggesting they are sensitive to the input from both languages.
  • Decreasing Influence with Age: As children get older and gain more mastery of their native language, the direct correlation between input frequency and production accuracy tends to decrease, though it remains significant.

Key Results

  • Word Repetition: For French-acquiring children aged 2-3, correct production scores for word-initial consonants were highly correlated with their input frequency (Pearson correlation: 0.81). This correlation decreased with age. For Drehu-acquiring children aged 3, the correlation was also significant (0.67).
  • Non-word Repetition:
    • Overall correct production scores for consonants increased with age from 3 to 5 years for both groups.
  • For phonemes that are rare in French but frequent in Drehu (/tf/ and /d3/), bilingual children had significantly better scores than monolingual French children.
  • For vowels, French monolinguals showed better production of /A/ and /i/ at certain ages compared to bilinguals, potentially linked to their specific input frequencies.

Limitations and Future Work

  • Age Range: The non-word repetition task was not administered to 2-year-old French children due to their inability to perform the task, limiting direct comparison at that youngest age for this specific task.
  • Vowel Context: The authors note that the four vowel groups in the French word repetition task were not always preceded by consonants of similar frequency, which might have influenced the results for vowels like /A/.
  • Future Work: The study suggests that further cross-linguistic comparisons are needed to fully understand the interplay between universal constraints and language-specific influences on phonological development. Investigating the impact of prosody on phoneme acquisition, as hinted at by previous research, could also be a fruitful avenue.

Summary updated

BibTeX
@inproceedings{monnin2010language,
  author = {Julia Monnin and Hélène Lœvenbruck},
  title = {{Language-specific influence on phoneme development: French and drehu data}},
  booktitle = {Interspeech},
  year = {2010},
}
Figure