Goldschneider, J. M. and R. M. Dekeyser. 2001: Explaining the 'natural order of L2 morpheme
acquisition' in English: a meta-analysis of multiple determinants. Language Learning 51, 1: 1-50.
Summary prepared by Florencia Franceschina
for the SLARG meeting of 8 May 2003
Main features of the study
Investigation of what determines orders of acquisition in L2 English
Based on pooled data from 12 primary studies (selected from an initial set of 25 studies) from
Includes pooled data from over 900 subjects.
The subject pool includes adults and children.
Includes pooled data from 28-29 L1s.
Seminal studies of L1 acquisition orders
DeVilliers and DeVilliers (1973)
Seminal studies of L2 acquisition orders
Dulay and Burt (1973) – L1 Spanish/L2 English children
Dulay and Burt (1974) – L1 Spanish and L1 Chinese / L2 English children
Bailey, Madden and Krashen (1974) – L1 Spanish and L1 various / L2 English adults
Larsen-Freeman (1975) – L1 various / L2 English adults
Pica (1983) – instructed vs naturalistic
Some accounts of L2 acquisition orders
Frequency in the input (Larsen-Freeman, 1975, 1976)
Frequency, perceptual salience and L1 (Mace-Matluck, 1977)
Gradual overcoming of speech-processing constraints (Pieneman and Johnston, 1987; Pieneman,
Qualities inherent to the functors (J.D. Brown, 1983) Favoured by the authors
Potential determinants investigated
G&DeK investigate whether acquisition orders are determined by a combination of the following factors
1. Perceptual salience
a. phonetic substance (number of phones)
b. syllabicity (presence/absence of a vowel in the surface form)
(All six functors were unstressed and all but articles came at the end of the word)
2. Semantic complexity
3. Morphophonological regularity
a. number of phonological alternations
b. homophony with other grammatical functors
4. Syntactic category
(The authors would also have liked to consider L1 transfer, but it was not done for practical reasons)
Criteria used to select studies for inclusion in this meta-analysis
1. L2 English only
2. Grammatical functors only
3. ESL, but not EFL
4. Oral production data only
5. Children and adults
6. Ignored type of instruction or exposure
7. LOR should not be too high
8. SOC as a scoring system
9. Availability of percentage of correct SOC scores for each functor
10. Restricted to the 6 more commonly found functors across studies:
- present progressive -ing
- plural -s
- possessive 's
- articles a, an, the
- 3PS present -s
- regular past -ed
A large proportion of the total variance in acquisition order is explained by a combination of the five
“The hypothesis that a large portion of the variance is explained by these five factors is supported”
G&DeK conclude that the data support a connectionist view of language learning, based on
inductive learning rather than guided by innate principles.
Was the operationalization of determinants adequate?
Was the claim that all the factors investigated are some form of salience justified?
Can we conclude that something is acquired by only looking at spontaneous production data?
How can this account help us to understand more advanced L2 development?
Can this account help us predict detailed paths of development beyond morphemes?
Are their pedagogical remarks at the end justified? (I think their claims would imply that
instructed learners should be better than people with naturalistic exposure. Is that really so?)
If we manipulated the input in a clever way, could we make things salient differently and change
the order of acquisition of a group of learners?