Measure of Cluster Proximity

The Measure of Cluster Proximity (MCP) is a quantitative metric to evaluate consonant cluster production in speech development. This measure addresses the issue of quantitatively differentiating stages in cluster production highlighted by the common patterns of cluster reduction, vowel epenthesis, two-member productions, with either substituted or correct phones.

References

Babatsouli, Elena & Dimitrios Sotiropoulos. 2018. A Measure for Cluster Proximity (mcp) in Child Speech. Clinical Linguistics & Phonetics 32(12). 1071–1089. https://doi.org/10.1080/02699206.2018.1510982.

Babatsouli, Elena (2021) Correlation between the Measure for Cluster Proximity (MCP) and the Percentage of Consonants Correct (PCC). Clinical Linguistics and Phonetics, 35(1), 65-83. doi: 10.1080/02699206.2020. 174418.

Data Preparation

Session data is queried by word for each IPA Target and IPA Actual tuple. Proper syllabification and phone alignment is required for accurate results. All participants should have unique names/identification numbers and a speaker should be assigned to all records.

MCP Data

The measure is computed on two-consonant clusters only (tri-consonantal or longer clusters are ignored by the algorithm). The MCP analysis will add the following information to each cluster identified in the query:
  1. Context
    Context will include the type of target cluster as well as the word location. Four types of clusters are considered in the analysis:
    1. Word-initial Onsets (OOi)
    2. Word-medial Onsets (OOm)
    3. Word-medial Coda+Onset (COm)
    4. Word-final Codas (CCf)
  2. Pattern

    Pattern is a two or three letter code describing the IPA Target and IPA Actual alignment. S represents substitution, D represents deletion, C represents correct, and V represents vowel epenthesis.

  3. Proximity

    Each Pattern is assigned a proximity value between 0% and 100%, with all values set in 12.5% increments between 0% and 100%, each of which provides a combined assessment of both the number of phones produced and the quality of these phones relative to the target cluster (Babatsouli & Sotiropoulos 2018; Babatsouli 2020), as shown in the proximity lookup table below.

    Table 1. Proximity Lookup Table
    Pattern Proximity (%)
    DD 0
    CC 100
    DS 12.5
    SD 12.5
    SVD 18.75
    DVS 18.75
    CD 25
    DC 25
    CVD 31.25
    DVC 31.25
    SVS 37.5
    CVS 50
    SVC 50
    CVC 62.5
    SS 75
    CS 87.5
    SC 87.5

Parameters

Figure 1. Measure of Cluster Proximity Parameters
Cluster types included in the analysis may be selected using checkboxes as shown above:
  1. Word-initial Onsets (OOi)
  2. Word-medial Onsets (OOm)
  3. Word-medial Coda+Onset (COm)
  4. Word-final Codas (CCf)

Report Outline

A sample table of contents is displayed below. Bold level elements are section headers while italic items are tables.

  • Measure of Cluster Proximity

    • Summary

    • Participant 1
      • Summary
      • Aggregate (all clusters)
      • Aggregate (OOi)
      • Aggregate (OOm)
      • Aggregate (COm)
      • Aggregate (CCf)
    • … (for each selected participant)

    • Listing