122
ACQ
Volume 13, Number 3 2011
ACQ
uiring Knowledge in Speech, Language and Hearing
community members in Canada: aim to involve primary
caregivers in assessment sessions; and begin sessions with
receptive language tasks (Ball & Bernhardt, 2008).
Collectively, these are not difficult steps for practising
clinicians to take. Time and efficiency are not valid
reasons to ignore these strategies since rejecting such
recommendations would potentially increase the likelihood
of obtaining invalid assessment data, in turn adding to
workloads unnecessarily. As a first step to culturally relevant
criterion referenced norms for AE speakers, Gould (2008b)
shows that there are effective, culturally relevant methods
of language sampling that result in valid, representative
discourse data. This non-standard approach to assessment
can be used in preference to formal, standardised
assessments which can culturally alienate and disempower
the child (e.g., through feelings of “shame”) as soon as he/
she walks into the unfamiliar clinic environment.
How then should valid language samples be analysed?
Given the current absence of norms for AE and Indigenous
Australian language speakers, language sample analyses
are limited by the fact that criterion-based measures such
as mean length of utterance (MLU) are based on Standard
English as a first language. Two studies discuss the
development of standard reference criteria for Indigenous
First Nation and Indigenous South African language
communities respectively (Ball & Bernhardt, 2008; Naudé,
Louw, & Weideman, 2007). Naudé et al. (2007) recognised
the demand for immediate, valid measures in multilingual,
English-dominant societies and thus explored the utility of
disregarding the advised method of testing development
in both/all languages (Speech Pathology Australia, 2009).
As an alternative, they observed typical development and
analysed language samples of the bi-/multilingual child’s
acquisition of English as an
additional
language. Criterion
referenced measures including MLU and type-token ratio
were used in analyses. This procedure acknowledged
Nicoladis and Genesée’s (1997) assertion that a valid
measure should come second to the establishment of
typical development. Naudé et al. (2007) aimed to describe
a typical English language profile for a selected group
of urban multilingual South African preschoolers. After
establishing the expected language patterns, clinicians
were able to use this set of indicators as a checklist
to determine deviations from expected performance.
Interestingly, wide ranges of MLU were described within
age-groups (potentially due to individual difference) and
thus the authors suggested that alternative measures to
help assess expressive language skill development should
be investigated.
The above findings indicate that there is potential for
the development of Indigenous linguistic, and/or dialectal,
developmental expectations and thus the establishment
of reliable criterion referenced measures. However, the
heterogeneity of Australia’s Indigenous population should
always be considered. There is no guarantee that the
validity of a particular method will not change between
different communities.
Dynamic assessment
Over the past 15 years dynamic assessment, a mediated
form of assessment, has received attention in the speech
pathology discipline as a means of assessing and providing
appropriate language intervention for CALD children (Carter
et al., 2005). Based on methods in educational psychology
(Ukrainetz, Harpell, Walsh, & Coyle, 2000), dynamic
Contextualised language sample analysis
The comprehensive assessment of any child’s
communication skills should include the analysis of a
spontaneous language sample. The sample provides the
SP with baseline, pre-intervention data and evidence of
discourse skills. A variety of methods are available to elicit
connected speech samples including free-play, activity-
based play, narratives, and conversation. In a culturally
different environment, young children are potentially reticent
to communicate naturally (Moses & Wigglesworth, 2008),
thus affecting the validity of the obtained sample.
Gould (2008b) showed that language sample analysis
has the potential to be a valid, culturally appropriate method
of assessment for AE-speaking children. In her longitudinal
study Gould (2008b) identified methods of language
sample elicitation that are more effective than others
when assessing language development in an AE speech
community in Queensland, Australia. The study considered
a number of different methods: a) minimally structured
storytelling (natural conversation and play); b) elicited story
generation (first person narratives and conversations about
local Aboriginal stories); c) story retelling (verbal narrative
reconstruction of a Western story [picture book style] and
an unfamiliar Aboriginal story [told on video by an Aboriginal
man]). Gould found that situational, environmental, and
linguistic contexts surrounding the language sample
collection affected the suitability of the sampling technique
(Gould, 2008b). The setting, topic, and interlocutors
involved were identified as the main variables to influence
the effectiveness of eliciting qualitatively valid and reliable
data. The most effective language samples were elicited
through conversations between the child and AE-speaking
adults from the child’s community. Audio-visual recordings
of free play (while speaking AE) were also regarded as
effective. Gould (2008b) further found that the most difficult
methods of elicitation for children aged 4–5 years included
story retelling and first person narratives with visual picture
cues. Finally, and most pertinently, general conversation
with a non-Aboriginal adult was not generally regarded
as a reliable method of eliciting valid spoken language
samples. The less useful strategies resulted in increased
amount of effort by the examiner to elicit spoken language
and a reduction in sentence length and complexity of
elicited utterances. Of particular note was the reduction of
spontaneous communication by Aboriginal children when
retelling western stories. More representative samples
were elicited when children were exposed to an unfamiliar
Aboriginal story. In addition to highlighting language
production differences between elicitation contexts, these
results remind us of the potential underlying cross-cultural
variations that exist for discourse and narrative structures
(see Kaplan, 1972, for an introduction) warning of the
limitations of standardised tests using story retell.
Gould (2008c) provides a number of considerations to
maximise cultural appropriateness during assessment trials,
including the benefit of becoming familiar with examinees
prior to assessments; the use of culturally meaningful
language during sessions; avoidance of the “pull-out”
method where possible to limit feelings of “shame”; informal
assessment to minimise potential power imbalances
between SAE and AE speakers, and; the need to consult
and collaborate with Indigenous people about appropriate
cultural interactions and expectations. Two additional
considerations have been identified by First Nations